Quantitative MRI of parallel changes of articular cartilage and underlying trabecular bone in degeneration

OsteoArthritis and Cartilage (2007) 15, 1149e1157 ª 2007 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.joca.2007.03.019

International Cartilage Repair Society

Quantitative MRI of parallel changes of articular cartilage and underlying trabecular bone in degeneration E. Lammentausta M.Sc.yz*, P. Kiviranta B.M.y, J. To¨yra¨s Ph.D.x, M. M. Hyttinen M.D.k, I. Kiviranta M.D., Ph.D.{, M. T. Nieminen Ph.D.z and J. S. Jurvelin Ph.D.y# y Department of Physics, University of Kuopio, Kuopio, Finland z Department of Diagnostic Radiology, Oulu University Hospital, Oulu, Finland x Department of Clinical Neurophysiology, Kuopio University Hospital, Kuopio, Finland k Institute of Biomedicine, Anatomy, University of Kuopio, Kuopio, Finland { Department of Orthopaedics and Traumatology, Jyva¨skyla¨ Central Hospital, Jyva¨skyla¨, Finland # Department of Clinical Physiology and Nuclear Medicine, Kuopio University Hospital, Kuopio, Finland Summary Objective: To determine the interrelations between degenerative changes in articular cartilage and underlying trabecular bone during development of osteoarthritis and to test the ability of quantitative magnetic resonance imaging (MRI) to detect those changes. Methods: Human cadaver patellae were investigated with quantitative MRI methods, T2 and dGEMRIC, at 1.5 T. Same measurements for isolated cartilage samples were performed at 9.4 T. Bone samples, taken at sites matched with cartilage analyses, were measured with MRI and peripheral quantitative computed tomography (pQCT). Mechanical and quantitative microscopic methods were also utilized for both cartilage and bone samples. Results: Significant differences were found between the samples with different stages of degeneration in mechanical properties, T2 at 1.5 T and proteoglycan (PG) content of articular cartilage. dGEMRIC at 9.4 T discerned samples with advanced degeneration from the others. Bone variables measured with pQCT discerned samples with no or minimal and advanced degeneration, and mechanical properties of trabecular bone discerned samples with no or minimal degeneration from the others. Significant linear correlations were found between the bone and cartilage parameters. Characteristically, associations between variables were stronger within the samples with no or minimal degeneration compared to all samples. Conclusions: Quantitative MRI variables, especially T2 relaxation time of articular cartilage, may be feasible surrogate markers for early and advanced osteoarthritic changes in joint tissues, including decreased elastic moduli, PG and collagen contents of cartilage and mineral density and volume fraction of trabecular bone. Further work is required to resolve the relaxation mechanisms at clinically applicable field strengths. ª 2007 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. Key words: Articular cartilage, Trabecular bone, Human, MRI, Biomechanics, BMD.

plate increases and sclerotic increase of the amount of the trabeculae may appear beneath it. However, the volumetric mineral density of trabecular bone decreases due to incomplete mineralization of trabecular structure2. Cyst-like bone cavities may also appear. At final stages of osteoarthritis, cartilage is completely worn out leaving the dense subchondral plate functioning as the articulating surface. These changes progressively compromise the mechanical function of joint. As the early stages of the disease are practically asymptomatic the time sequence of the initial changes in cartilage and bone remains poorly understood. Recent in vivo studies, focusing on both articular cartilage and trabecular bone, utilize measurement of joint space narrowing as a diagnostic indicator of osteoarthritis3,4. Unfortunately, serious degeneration of joint tissues has already begun before any changes in cartilage thickness or joint space can be detected. Quantitative magnetic resonance imaging (qMRI) parameters could serve as early markers of joint degeneration and changes in the mechanical properties of cartilage and underlying bone. Several qMRI techniques have recently been introduced for the noninvasive assessment of structure and composition of articular

Introduction The living joint is a mechanical unit with several interacting components, such as cartilage, bone, ligament and tendon. Due to continuous mechanical interaction between joint structures pathological alterations in one component are likely to induce either compensative or pathological changes in other components. In the first stage of the most common degenerative joint disease, osteoarthritis, degenerative changes in both articular cartilage and trabecular bone take place. For articular cartilage, among the first changes are increase in water content, loss of proteoglycans (PGs) and disruption of collagen fibrils in superficial cartilage, gradually advancing and spreading to deeper tissue and eventually leading to complete disorganization of the cartilage matrix1. For trabecular bone, density of the subchondral *Address correspondence and reprint requests to: Eveliina Lammentausta, M.Sc., Department of Diagnostic Radiology, Oulu University Hospital, POB 50, FI-90029 OYS, Oulu, Finland. Tel: 358-8-3152480; Fax: 358-8-3155420; E-mail: eveliina.lammen [email protected], http://www.luotain.uku.fi Received 21 December 2006; revision accepted 27 March 2007.

1149

1150

E. Lammentausta et al.: qMRI of parallel cartilage and bone changes

cartilage and trabecular bone5,6. T2 relaxation time of articular cartilage has been found to be sensitive to the orientation and amount of collagen fibrils7e9, while T1 relaxation time in presence of an ionic contrast agent (delayed Gd-Enhanced MRI of Cartilage (dGEMRIC) technique) reflects the PG distribution in cartilage tissue10,11. For trabecular bone, T2* relaxation time is associated with bone mineral density (BMD)12e14, and trabecular structure can be assessed morphometrically after binarizing MRI images15e17. The aim of the present study was to determine the interrelations of changes in articular cartilage and underlying trabecular bone at different stages of osteoarthritic degeneration. Particularly, we tested the ability of clinical MRI techniques to reflect these changes and their interrelations during progressive degeneration. MRI variables obtained with clinical 1.5 T scanner and MRI variables of cartilage obtained at 9.4 T research magnet and BMD as measured with peripheral quantitative computed tomography (pQCT) were compared to mechanical properties measured separately for bone and cartilage samples. Additional reference measurements were conducted using quantitative microscopic techniques, including Fourier transform infrared spectroscopy imaging (FTIRI) to estimate the collagen content of cartilage, and optical density measurements of cartilage PGs, and polarized light microscopy of bone to estimate the bone volume fraction. The data are partially based on two earlier publications that assessed cartilage18 and bone properties19 and thereby help to draw new important conclusions on the interaction of these tissues in degenerative joint disease.

Methods SAMPLE PREPARATION

Human cadaver knee joints (N ¼ 14, 12 males, 2 females, mean age ¼ 55  18 years) were obtained within 48 h postmortem from the Jyva¨skyla¨ Central Hospital, Jyva¨skyla¨, Finland, with permission from the National Authority of Medicolegal Affairs, Helsinki, Finland (permission 1781/32/ 200/01), and dissected bones were frozen separately. Healthy joints as well as joints with different degrees of degeneration were accepted for the study. The right patellae were thawed overnight, and six topographical locations were assessed to cover the entire articular surface (Fig. 1). For the reproducible localization the sites were marked on the cartilage surface using a fine felt-tipped pen. Intact patellae were used for 1.5 T MRI and pQCT measurements. For 9.4 T MRI and mechanical measurements of cartilage, 4-mm cylindrical samples were used. MRI, pQCT and mechanical measurement methods have been described in more detail earlier18,19. All data analyses were conducted using in-house MatLab scripts (MatLab, Mathworks Inc., Natick, MA, USA). Abbreviations for all measured and calculated parameters are shown in Table I. MRI MEASUREMENTS OF CARTILAGE AND BONE

Patellae were equilibrated overnight in 0.5 mM Gadoliniumdiethylene triamide pentaacetic acid (Gd-DTPA2) solution (Magnevist, Schering AG, Germany) and frozen. It has been shown that small Gd-DTPA2 concentrations, such as that applied here, have no significant effect on T2 relaxation time of cartilage20. The present Gd-concentration was chosen as, based on measurements of T1 in the presence and absence of Gd-DTPA2, the in vivo concentration has been estimated to be less than 1 mM20. However, assessment of Gd-DTPA2 concentration in vivo

Fig. 1. Test sites for articular cartilage and trabecular bone. SM ¼ superomedial, SL ¼ superolateral, CM ¼ central medial, CL ¼ central lateral, IM ¼ inferomedial, and IL ¼ inferolateral.

is complicated because the exact relaxivity, i.e., the ability of the contrast agent to enhance relaxation, of cartilage is not known in vivo10. Prior to measurements patellae were thawed, wrapped in plastic film to prevent dehydration, and MRI-visible capsules containing peanut oil were attached at the measurement sites to serve as localization markers (Vitol, Cardinal Health UK 414 Ltd, Wiltshire, Great Britain). A clinical 1.5 T scanner (Signa TwinSpeed, GE Healthcare, Milwaukee, WI, USA) was used together with a 300 receiving surface coil and the body coil as the transmitting coil. The articular surface of intact patellae was oriented parallel to the B0 field to emulate clinical patient positioning. For T2 mapping a series of eight images were acquired using a multi-slice multi-echo spin echo sequence with improved slice profile (repetition time TR ¼ 1000 ms, echo Table I Abbreviations of measured and calculated parameters dGEMRICs [ms] dGEMRIC [ms] T2,s [ms] T2 [ms] T2* [ms] BV/TVSE [%] BV/TVGE [%] BMD [mg/cm3] BV/TVpQCT [%] BV/TVREF [%] Es,c [MPa] Ed,c [MPa] Es,b [MPa] sy [MPa] su [MPa] PG [OD] MS Collagen [absorption]

Gd-enhanced T1 relaxation time of superficial articular cartilage Gd-enhanced T1 relaxation time of articular cartilage T2 relaxation time of superficial articular cartilage T2 relaxation time of articular cartilage T2* relaxation time of trabecular bone Bone volume fraction assessed from spin echo measurements Bone volume fraction assessed from gradient echo measurements BMD measured with pQCT Bone volume fraction measured with pQCT Bone volume fraction measured with polarized light microscopy Elastic modulus of articular cartilage Dynamic modulus of articular cartilage Elastic modulus of trabecular bone Yield stress of trabecular bone Ultimate strength of trabecular bone PG content of articular cartilage MS of articular cartilage Collagen content of articular cartilage

1151

Osteoarthritis and Cartilage Vol. 15, No. 10 time TE ¼ 10.3e82.4 ms, echo train length ETL ¼ 8, 3-mm slice thickness, a field-of-view of 8 cm and 256  256 matrix size to yield 0.313 mm pixel size, measured at room temperature). This was followed by a T1 relaxation time (dGEMRIC) measurement (single-slice inversion recovery fast spin echo sequence, TR ¼ 1700 ms, TE ¼ 11 ms, six inversion times (TI’s) between 50 and 1600 ms, ETL ¼ 6), and T2* measurement of bone (gradient echo TR ¼ 100 ms, six TE’s between 4.7 and 28 ms, flip angle ¼ 30 ). Subsequently, full-thickness 4-mm cartilage disks were prepared from each measurement site using a biopsy punch and a razor blade. The topographical locations of extracted samples were marked on the bone surface with a felt-tipped pen, and the patellae were frozen. Patellar bones were thawed before fast gradient echo sequence measurements (TR ¼ 30 ms, TE ¼ 4 ms, flip angle ¼ 40 , 1-mm slice thickness, a field-of-view of 6 cm and 256  256 matrix yielding 0.234 mm pixel size, measured at room temperature) were used. From each patella, 40 consecutive slices were imaged to cover the entire trabecular bone volume for structural assessment. Patellae were frozen for further experiments. For 9.4 T measurements of cartilage samples, an Oxford NMR vertical magnet (Oxford Instruments PLC, Witney, UK), a SMIS console (SMIS Ltd, Surrey, UK), and a 5-mm high-resolution volume spectroscopy probe (Varian Associates Inc., Palo Alto, CA, USA) were used. Prior to measurements, the cylindrical samples were thawed and sealed in a test tube (dia. ¼ 5 mm) immersed in Gd-DTPA2 solution to maintain the contrast agent concentration. The samples were located axially in the center of the coil and the sample surface was oriented perpendicular to the B0 field, as limited by the coil construction. T2 and T1 (dGEMRIC) were determined using a spin echo sequence (TR ¼ 1500 ms, TE ¼ 14e80 ms, 1-mm slice thickness, 0.039 mm resolution across cartilage depth, at 25  1 C) and saturation recovery sequence (TE ¼ 14 ms, six TR’s between 100 and 2000 ms), respectively.

Fig. 2. Localization of the sample sites for articular cartilage (black rectangles) and trabecular bone (white squares). Two locations were marked at each slice, one on medial and one on lateral facet. Capsules serving as localization markers and spin density reference were attached on a plastic film wrapped firmly around the patella. The ROI were chosen to be rectangular and approximately 10 pixels (3.125 mm) wide for cartilage and 23 pixels (7.19 mm) wide for bone.

in selected ROI and localization marker functioning also as oil reference22,23. Trabeculae were assumed to have negligible MRI signal; thus the spin density of each ROI was assumed to be entirely due to bone marrow inside the ROI. The oil reference was chosen to represent the case of ROI consisting of only bone marrow. The bone marrow volume fraction in each ROI was calculated by dividing the signal intensity of each ROI with the spin density of the oil reference in each image. Bone volume fraction (BV/ TVSE) was calculated by subtracting the bone marrow volume fraction from unity. The markers were vitamin capsules with peanut oil as an auxiliary substance (Vitol, Cardinal Health UK 414 Ltd, Wiltshire, UK).

MRI ANALYSES

The MRI parameter maps were obtained by fitting data into exponential relaxation equations assuming a monoexponential decay21. Regions of interest (ROI) with a width matching the slice thickness were manually segmented at the measurement sites as marked by the MRI-visible capsules (Fig. 2). To avoid partial volume effects, the most superficial pixel was omitted. Relaxation times for the approximately 1 mm of the most superficial tissue (three pixels at 1.5 T and eight pixels at 9.4 T) as well as bulk values covering the full thickness uncalcified cartilage were determined. As the thickness of the actual superficial cartilage layer is beyond the resolution of 1.5 T measurements, the superficial values of MR parameters refer to thicker region than the histological superficial cartilage layer as determined by the collagen fibril orientation. For bone analysis, 7  7 mm2 ROI were segmented manually under subchondral plate at the measurement sites (Fig. 2). From structural images a 7  7  7 mm volume of interest (VOI) was assessed, having the centermost slice at the location of the T2* measurements. The apparent bone volume fraction, i.e., the ratio of the bone volume to the total tissue volume, (BV/TVGE) was calculated based on the histograms of signal intensities in the selected volumes16. Spin echo measurements used for T2 calculations were also used to estimate apparent bone volume fraction (BV/TVSE) by comparing the spin density of bone marrow

pQCT MEASUREMENTS

Prior to pQCT measurements, patellae were thawed in saline immersion. Plastic markers were used to localize the six topographical locations in each patella. Volumetric BMD was measured with localization corresponding to the T2* measurements (58 kV, 0.175 mA, nominal in-plane resolution 0.200 mm, 0.5-mm slice thickness, XCT 2000, Stratec, Birkenfeld, Germany). A BMD value for a 7  7 mm ROI, localized into the trabecular bone under dense subchondral plate, was calculated. Apparent bone area fraction was determined for each ROI by using the same histogram based technique as for the MRI gradient echo images and used as an estimate of bone volume fraction (BV/TVpQCT) in 3-D space. BIOMECHANICAL TESTING OF ARTICULAR CARTILAGE

Biomechanical testing was performed with a custommade high-resolution material testing device [load cell with resolution of 5 mN (Honeywell Sensotec, Columbus, OH, USA) and a precision motion controller with resolution of 0.1 mm (Newport, Irvine, CA, USA)]24. A stress-relaxation test was performed in unconfined compression geometry. After establishing a proper surface contact, a 10% compressive pre-strain was applied followed by 1-h relaxation.

1152

E. Lammentausta et al.: qMRI of parallel cartilage and bone changes

Subsequently, this was followed by three 2% compressive steps at 1 mm/s ramp velocity and 40 min relaxation after each step. Young’s modulus (Es,c) was determined from the equilibrium response. A dynamic test with 1 Hz frequency and 1% strain amplitude was conducted after the stress-relaxation test, and the dynamic modulus (Ed,c) was determined as the stress/strain ratio of peak-to-peak values25,26. BIOMECHANICAL TESTING OF TRABECULAR BONE

Cylindrical samples (dia. ¼ 7 mm, length ¼ 7 mm) were isolated from the sites of interest. A grinding system (Macro Exakt 310 CP, Exakt, Hamburg, Germany) was used to parallelize the ends of cylindrical plugs and remove the dense subchondral plate. The destructive testing protocol included 2-min equilibration of the sample under a pre-stress of 10 N, five non-destructive dynamic cycles (0.5% strain) and destructive compression to 5% strain with a ramp velocity of 5 mm/s (Instron FastTrack 8874, Instron, Norwood, MA, USA). Young’s modulus (Es,b) was determined as the slope of the linear fit to the stress/strain curve between 40% and 65% of the maximum stress27. An offset method with strain offset of 0.03% was used to determine the yield point and yield stress (sy)28. The ultimate strength (su) was obtained as the maximum stress. HISTOLOGICAL ANALYSES

The PG content of matrix was assessed with digital densitometry of 3-mm thick Safranin-O stained sections29. Collagen content was estimated from unstained sections with Fourier transformed infrared imaging technique30. The histological Mankin score (MS) of the samples was evaluated by three of the authors independently from blind-coded Safranin-O stained sections31. Since the samples were detached from subchondral bone, the integrity of the tidemark could not be evaluated. Bone cylinders too short for biomechanical testing (specimen length less than 7 mm, see below) were prepared for polarized light microscopy. Specimens were decalcified in ethylenediaminetetraacetic acid (EDTA)eformalin, dehydrated in ethanol solutions, and embedded in paraffin as previously described29. Histological sections (7 mm thick) were cut with an LKB Historange microtome (LKB, Bromma, Sweden). After dewaxing and rehydration of the sections, PGs were digested with testicular hyaluronidase32, and unstained sections were prepared for polarized light microscopy. Collagen network of the bone matrix was visualized with crossed polarizers and strain-free optics. Orientation independent birefringence image of the organized collagen matrix was generated after stepwise rotation of the polarizereanalyzer pair33 and calculation of the theoretical intensity maximum for every pixel in the image. The method was totally insensitive to cells as well as to loose connective and adipose tissue in the bone marrow space since these structures do not show any regular organization of fibrillar collagen. The areal fraction of the bone collagen matrix was calculated from the binarized, background corrected birefringence images using 9.26 mm spatial pixel resolution, and used as an estimator of bone volume in the 3-D space (BV/TVREF). A total of 84 bone and cartilage samples were measured and analyzed with 1.5 T MRI and pQCT methods. The total number of cartilage samples measured successfully with 9.4 T MRI, biomechanical testing and microscopic methods

was 75. Some cartilage samples were unsuccessfully isolated from the bone and some disks with non-uniform thickness could not be mechanically tested. The cylindrical bone plug required for mechanical testing was not detachable from all sites of interest due to insufficient thickness of trabecular bone. Additionally, the curvature of the patellar surface caused overlapping of adjacent medial and lateral plugs from which other had to be rejected (each plug was drilled at perpendicular direction to the articular surface). Therefore mechanical testing was performed for 46 samples. Twenty-four samples too short for mechanical testing were analyzed microscopically. The samples were divided into three groups with different degree of degeneration according to their MS: samples with no or minimal degeneration (group I, MS < 4), moderate degeneration (group II, 4  MS < 8) and advanced degeneration (group III, MS  8) (Table II). The number of samples was different for bone and cartilage samples because of the aforementioned technical difficulties in bone sample preparation. STATISTICAL ANALYSES

For comparison of measured variables, linear Pearson correlation coefficients were determined separately for all samples and for samples with no or minimal degeneration. The statistical differences between different stages of degeneration were tested using KruskaleWallis and Manne Whitney U tests. Statistical analyses were conducted using SPSS software (version 11.5, SPSS Inc., Chicago, IL, USA). The significance of the difference between two correlation coefficients was tested by using the standard Fisher Z-score test34. The significance of the difference between two correlation coefficients depends on the magnitude of the coefficient in question and the number of data points used for its calculation.

Results Both bulk and superficial dGEMRIC at 1.5 T and 9.4 T showed a similar trend of decreased values in group III indicating loss of PGs (Fig. 3). The difference was statistically significant at 9.4 T results but the large variation of measured values yields insignificant differences between the groups at 1.5 T. There was an increasing trend in T2 relaxation time along the progression of cartilage degeneration. Superficial T2 relaxation time at 1.5 T showed statistically significant differences between group III and other groups, and bulk T2 discerned even groups I and II. There were no significant differences in T2 between groups at 9.4 T; however a trend with slightly increasing values toward the advanced degeneration was shown. Mechanical properties

Table II Number of samples in different experimental groups. The samples were divided into groups according to their MS (MS < 4, no or minimal degeneration; 4  MS < 8, moderate degeneration; MS  8, advanced degeneration). Number of samples measured using different experimental techniques varies due to technical limitations in sample preparation MS < 4

4  MS < 8

MS  8

Cartilage

31

35

11

Bone Mechanical Histological

20 6

19 11

3 7

Group

1153

Osteoarthritis and Cartilage Vol. 15, No. 10

700 650 600 550 500 450

Es,c [MPa]

0.6 0.4 0.2 0

60

T2 [ms], 1.5 T

T2,s [ms], 1.5 T

70

55 50 45 40

50

900 850 800 750 700 650 600 550 500

52 50 48 46 44 42 40 38 36

9 8 7 6 5 4 3 2 1 0

2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4

1.2 0.8

280 260

1.4 1

300

80

T2 [ms], 9.4 T

750

320

90

42 40 38 36 34 32 30 28 26 24 0.32

Collagen [abs. unit]

260

100 340

T2,s [ms], 9.4 T

280

360

PG [OD]

300

dGEMRIC [ms], 1.5 T

320

Advanced degeneration

dGEMRIC [ms], 9.4 T

800

340

Moderate degeneration

Ed,c [MPa]

dGEMRICs [ms], 1.5 T

360

dGEMRICs [ms], 9.4 T

No/minimal degeneration

0.3 0.28 0.26 0.24 0.22 0.2 0.18

Fig. 3. Mean  SD values for cartilage variables divided into three groups according to the MS: group I (samples with no or minimal degeneration, the MS < 4), group II (moderate degeneration, 4  MS < 8 and group III (advanced degeneration, MS  8). The statistically significant differences (P < 0.05) between groups are denoted with brackets. The number of samples in groups measured with different modalities is shown in Table I.

and PG content as measured by optical density discerned all groups, and collagen content discerned group III from other groups, all three showing a systematic trend with decreasing values with advancing degeneration. For bone variables, BMD and BV/TV measured by pQCT showed significant difference between groups I and III with decreasing values along cartilage degeneration (Fig. 4). The mechanical properties of trabecular bone showed significant difference between groups I and II but not between groups I and III despite a similar mean values in groups II and III. This is probably due to small amount of samples in group III (n ¼ 3). MRI variables of trabecular bone did not show significant differences between groups divided by degree of cartilage degeneration. There were no statistically significant differences between samples with MS 0 (n ¼ 4) and all other samples in any variable except for PG content. The linear correlation coefficients between MRI variables of cartilage and mechanical and histological variables varied differently at different field strengths (Table III). The difference in correlation coefficients at different field strengths was statistically significant between superficial T2 and PG content (P ¼ 0.06), superficial T2 and Es,c (P ¼ 0.02) and between superficial T2 and Ed,c (P < 0.01). PG content correlated significantly with Es,c and Ed,c (r ¼ 0.55 and 0.62, respectively, P < 0.01). Collagen content correlated similarly with Es,c and Ed,c (r ¼ 0.58 and 0.68, respectively, P < 0.01). There were significant correlations also between mechanical properties and MRI and pQCT variables of trabecular bone (Table IV). The mechanical properties of cartilage and bone did not correlate significantly with each other despite similar trend along degeneration. T2* of

trabecular bone correlated positively with dGEMRIC at both field strengths (r ¼ 0.32, P < 0.01 and r ¼ 0.27, P < 0.05 for deep and superficial cartilage at 1.5 T, respectively, and r ¼ 0.46, P < 0.01 and r ¼ 0.44, P < 0.01 for deep and superficial cartilage at 9.4 T, respectively). T2* of bone correlated negatively with superficial T2 of cartilage at 1.5 T (r ¼ 0.31, P < 0.01). Superficial dGEMRIC at 1.5 T also correlated negatively with BMD (r ¼ 0.32, P < 0.01). Some of the linear correlation coefficients were higher when only samples with no or minimal degeneration were separately assessed. Despite the small number of samples, the increase of the correlation coefficient between bone volume fractions as measured with pQCT and microscopy was statistically significant (P < 0.05) (Table V). The increase of correlation coefficients between BV/TVREF and T2* and between BV/TVREF and BMD is illustrated in Fig. 5. Correlation coefficients between cartilage variables did not show similar increase as that for the correlation coefficients between bone and cartilage variables (Table VI). The increase was statistically significant in correlation coefficients between superficial dGEMRIC at 1.5 T and BMD (P < 0.01), Young’s modulus of cartilage and BMD (P < 0.05) and Young’s modulus and T2* (P < 0.05).

Discussion Human cadaver patellae were investigated with quantitative MRI and several reference methods, such as mechanical testing, microscopic analyses and pQCT measurements. Variables previously related to structural and mechanical

1154

E. Lammentausta et al.: qMRI of parallel cartilage and bone changes

0.35

0.5

0.3

8

550

7.5

500

7 6.5 6

0.25

0.4

0.4

1000

5.5

0.3 0.25

800 700 600 500 400

0.2

400 350

18

σy [MPa]

Es,b [MPa]

0.35

450

300

14

900

BV/TVpQCT

Advanced degeneration

BMD [mg/cm3]

0.4

T2* [ms]

0.6

BV/TVSE

BV/TVGE

0.45

Moderate degeneration

12 10 8 6

300

16

σu [MPa]

No/minimal degeneration

14 12 10 8 6

Fig. 4. Mean  SD values for bone variables divided into three groups according to the MS: group I (samples with no or minimal degeneration, the MS < 4), group II (moderate degeneration, 4  MS < 8 and group III (advanced degeneration, MS  8). The statistically significant differences (P < 0.05) between groups are denoted with brackets. The number of samples in groups measured with different modalities is shown in Table I.

properties of articular cartilage and trabecular bone were related to each other. Moreover, the degenerative status of the cartilage layer was determined histologically using MS and the changes in bone and cartilage as well as the changes in the relations between them were investigated as function of progression of degeneration. The present results from extensive analyses with several quantitative methods highlight the parallel changes in cartilage and underlying bone during degenerative process. Most of the revealed changes in tissue could be detected with MRI. At 1.5 T, T2 relaxation time of cartilage showed a significant increase with degeneration, whereas T2 measured at 9.4 T showed a similar trend, however, statistically insignificant. On the other hand, dGEMRIC at both field strengths showed a decreasing trend with degeneration but changes were significant at 9.4 T only. Increased T2 values have been earlier related to advanced degeneration in human cartilage35,36, however a T2 decrease has been reported in a study utilizing enzymatic digestion37. As compared to any digestion model selectively affecting specific components, spontaneous human OA can be considered a more complicated process, involving interaction of the cartilage Table III Linear correlation coefficients between MRI variables and mechanical properties of articular cartilage. There are significant differences in correlation coefficients of MRI parameters measured at different field strengths PG

Collagen

Es,c

Ed,c

1.5 T T2,s T2 dGEMRICs dGEMRIC

0.53** 0.39** NS NS

0.57** 0.43** NS NS

0.54** 0.40** 0.30** NS

0.62** 0.46** NS NS

9.4 T T2,s T2 dGEMRICs dGEMRIC

0.28* 0.47** 0.41** 0.45**

0.54** 0.50** 0.32** 0.30**

NS NS 0.35** 0.31**

0.24* 0.28* 0.39** 0.35**

**P < 0.01; *P < 0.05.

components1. The results from these earlier studies suggest that the relaxation mechanisms in articular cartilage are complicated and may be different at different field strengths, warranting further studies. T2 has been reported to be sensitive also to PG changes in porcine articular cartilage at 2.35 T with articular surface oriented along the direction of the B0 field38. When applying this orientation, the collagen fibers of deep layer of articular cartilage, the thickest layer, are perpendicular to the B0 field which considerably diminishes the nuclear dipolar interaction compared to the orientation with articular surface perpendicular to the B0 field, causing an increase in the signal detected from the deep cartilage39. Such an orientation may lead to a diminished contribution of collagen-related relaxation mechanisms and other, e.g., PG-related mechanisms may play a more important role in the deep cartilage. This may also partly explain the differences between T2 measured at different field strengths. The orientation dependency of T1 relaxation mechanisms is negligible40, whereas T2 depends strongly on the collagen orientation due to the dipolar interaction39,41. Hence, this might partially explain the significant connection between T2 and PG content. On the other hand, T2 and PG content have shown to have no significant correlation after PG depletion in bovine patellae42 as well as significant difference before and after a slight or severe PG loss induced by digestion in rat patellae43. The sample orientation with respect to the B0 field was different in these studies, and the effect of the magic angle on the results is difficult to assess. Thus, further investigation is required. As far as the authors know the present study is the first histological verification of relationship of dGEMRIC and OA-related changes in human articular cartilage. Although dGEMRIC revealed a decreasing trend in T1 values along tissue degeneration these differences were not statistically significant between groups with no or minimal degeneration and moderate degeneration at either field strength. This suggests that dGEMRIC, unlike T2 relaxation time, may not be sensitive enough to detect the earliest degenerative changes. Interestingly, the histological results show that PG content is significantly different between all groups but the

1155

Osteoarthritis and Cartilage Vol. 15, No. 10 Table IV Linear correlation coefficients between MRI and pQCT variables and mechanical and histological properties of trabecular bone BV/TVGE BV/TVSE T2* BV/TVpQCT BMD

Es,b

sy

su

BV/TVpQCT

BMD

BV/TVREF

0.32* NS NS 0.36* 0.44**

0.40** NS 0.34* 0.44** 0.61**

0.46** NS 0.33* 0.49** 0.64**

0.51** 0.55** NS e 0.82**

0.56** 0.66** 0.31* 0.82** e

NS NS NS 0.59** 0.58**

**P < 0.01; *P < 0.05.

Table V Linear correlation coefficients between bone volume fractions measured with different techniques (MRI, pQCT and polarized light microscopy). The number of measured samples was 31 (MS < 4) and 84 (all samples) for MRI and pQCT and 6 (MS < 4) and 24 (all samples) for microscopy. The correlation coefficients with gray background are calculated including only samples with MS less than 4. The correlation coefficients between BV/TVpQCT and BV/TVREF and between BV/TVSE and BV/TVREF were higher within samples with MS < 4

BV/TVSE

BV/TVpQCT

BV/TVREF

BV/TVGE

-

0.479*

0.506**

0.112

BV/TVSE

0.487**

-

0.549**

BV/TVpQCT

0.443*

0.573*

-

BV/TVREF

0.086 0.673 0.966** Samples with Mankin score < 4

0.392 0.586** -

All samples

BV/TVGE

be a sufficient resolution for healthy samples but that the degeneration process may also affect this relationship. For BV/TVGE, the chemical shift effect may cause additional blurring to the original image. The slice thickness was 0.5 mm and 1 mm for pQCT and for MRI, respectively, which also certainly affects the results, smaller slice thickness providing more accurate estimate. To improve the feasibility of BV/TVSE, an oil phantom with composition similar to that of the bone marrow is required. The content of yellow marrow, that is known to increase with age, is another challenge for this method. Thus none of the techniques used in the present study can be considered as a gold standard without further improvements.

Mankin score < 4 4 ≤ Mankin score < 8 Mankin score ≥ 8 9

(a)

T2* [ms]

8 7 6 5 4 0.1

0.2

0.3

0.4

0.5

0.6

0.5

0.6

BV/TVREF 700

(b)

600

BMD [mg/cm3]

difference in collagen content is not significant between sample groups of no or minimal degeneration and moderate degeneration. The relation between the present MRI and histological results is in disagreement with the current understanding on the relation of the present qMRI techniques and cartilage composition and structure, and implies that the relaxation processes in cartilage are more complicated than generally understood. The correlation between MRI variables and mechanical properties in patellar cartilage is reported to differ from other articular surfaces in human knee44, thus further investigations are required to determine the relationship of dGEMRIC and joint degeneration in other articular surfaces. BMD decreased along with osteoarthritic degeneration of cartilage. This is contradictory to previous results from histological evaluation of human samples45,46. This may be due to different choice of ROI. In the present study the dense subchondral bone was omitted from analyzed regions since it does not provide any NMR signal, and this might have excluded possible sclerosis. Some samples showed sclerotic changes near the subchondral plate in pQCT images. The decrease in mineral density of trabecular bone may partially be explained by the age-related changes, such as osteoporosis. The age of joint donors was significantly lower within samples with no or minimal degeneration. Four different calculations were used for bone volume fraction. BV/TVREF was calculated from sections prepared for polarized light microscopy, and it was the only calculation with resolution sufficient for exact binarizing (9.26 mm for polarized light microscopy as compared to 200 mm for pQCT and 234 mm for MRI). However, this is a histological method not suitable for in vivo use. BV/TVpQCT and BV/ TVGE were based on the image histogram and provide reasonable estimates for images with better resolution. BV/ TVpQCT agreed well with BV/TVREF within samples with no or minimal degeneration, suggesting that 200 mm would

500 400 300 200 0.1

0.2

0.3

0.4

BV/TVREF Fig. 5. Scatter plots of T2* relaxation time and BMD as a function of microscopically measured apparent bone volume fraction (BV/ TVREF). In both cases the correlation coefficient is increased significantly when samples with MS < 4 only are included. In subfigure (a) the correlation coefficient increases from 0.09 (not significant) to 0.86 (P < 0.05). In subfigure (b) the coefficient changes from 0.58 (P < 0.01) to 0.91 (P < 0.01). In both cases the change of correlation coefficient is significant (P < 0.05). The line is fitted to the sample group with MS < 4.

1156

E. Lammentausta et al.: qMRI of parallel cartilage and bone changes

Cartilage variables

Table VI Linear correlation coefficients between cartilage and bone variables and the increase in coefficients when including only samples with MS less than 4. All correlation coefficients are higher for samples with MS < 4 than for all samples

dGEMRICs Es,c Ed

BMD

Bone variables σy T2*

-0.66**

0.48**

-0.46*

-0.47*

-0.32**

0.32**

NS

NS

-0.47** NS

0.70** 0.39**

σu

-0.37* 0.55** NS 0.33** Samples with Mankin score < 4 All samples

**P < 0.01; *P < 0.05.

Mechanical properties of articular cartilage and trabecular bone showed a similar trend toward smaller values along cartilage degeneration but there were no significant linear correlations between mechanical parameters of bone and cartilage. Yet, cartilage and bone MRI parameters were correlated. This suggests that the chain of degenerative changes is complex and needs further investigation. Some linear associations were higher when only the samples with no or minimal degeneration were included. This result, together with the results of earlier studies using enzymatic digestion, suggests that quantitative MR parameters reflect the mechanical properties of healthy tissue but the actual degeneration affects the tissue in a complex way. Additionally, there might be several types of subprocesses in OA development, and quantitative parameters which are sensitive to changes in single joint tissue component may not detect all degenerative changes. The current results are obtained using human patellae. Because of the specific loading conditions, the trabecular structure may be more adapted to tensile stress than compressive weight-bearing47. Similar adaptation may also occur in cartilage. The relation between MRI parameters and mechanical properties at patellar surface have been shown to differ from those at other articular surfaces of the human knee44. Thus, generalization of the present results for other joint surfaces may not be warranted. Based on the present results, the characteristic changes during osteoarthritis include decrease in elastic moduli of bone and cartilage, PG and collagen contents of cartilage and BMD and bone volume fraction of trabecular bone. Despite the modest linear relations between the MRI and reference parameters, current results suggest that parallel cartilage and bone changes in degenerative joint disease can be detected with qMRI techniques, particularly T2 of articular cartilage and T2* of trabecular bone. Even though previous and present results suggest that T2 is sensitive to more than one structural component of articular cartilage, it seems that T2 is also effective in discerning different stages of osteoarthritis.

Acknowledgements The authors would like to thank the Department of Pathology, Jyva¨skyla¨ Central Hospital, Jyva¨skyla¨, Finland, for

initial sample preparation. Financial support from the Academy of Finland (grant no. 205886), the Finnish Academy of Science and Letters, Vilho, Yrjo¨ and Kalle Va¨isa¨la¨ Foundation and the Finnish Cultural Foundation of Northern Savo is gratefully acknowledged.

References 1. Buckwalter JA, Mankin HJ. Articular cartilage, Part II: Degeneration and osteoarthrosis, repair, regeneration, and transplantation. J Bone Joint Surg Am 1997;79: 612e32. 2. Burr DB. Anatomy and physiology of the mineralized tissues: role in the pathogenesis of osteoarthrosis. Osteoarthritis Cartilage 2004;12(Suppl A):S20e30. 3. Messent EA, Ward RJ, Tonkin CJ, Buckland-Wright C. Tibial cancellous bone changes in patients with knee osteoarthritis. A short-term longitudinal study using Fractal Signature Analysis. Osteoarthritis Cartilage 2005;463e70. 4. Lindsey C, Narasimhan A, Adolfo J, Jin H, Steinbach L, Link T, et al. Magnetic resonance evaluation of the interrelationship between articular cartilage and trabecular bone on the osteoarthritic knee. Osteoarthritis Cartilage 2004;12:86e96. 5. Burstein D, Gray M. New MRI techniques for imaging cartilage. J Bone Joint Surg Am 2003;85-A:70e7. 6. Wehrli FW, Saha PK, Gomberg BR, Song HK, Snyder PJ, Benito M, et al. Role of magnetic resonance for assessing structure and function of trabecular bone. Top Magn Reson Imaging 2002;13: 335e56. 7. Nieminen MT, To¨yra¨s J, Rieppo J, Hakuma¨ki JM, Silvennoinen J, Helminen HJ, et al. Quantitative MR microscopy of enzymatically degraded articular cartilage. Magn Reson Med 2000;43:676e81. 8. Xia Y, Moody JB, Burton-Wurster N, Lust G. Quantitative in situ correlation between microscopic MRI and polarized light microscopy studies of articular cartilage. Osteoarthritis Cartilage 2001;9:393e406. 9. Nieminen MT, Rieppo J, To¨yra¨s J, Hakuma¨ki JM, Silvennoinen MJ, Hyttinen MM, et al. T2 relaxation reveals spatial collagen architecture in articular cartilage: a comparative quantitative MRI and polarized light microscopic study. Magn Reson Med 2001;46: 487e93. 10. Bashir A, Gray ML, Hartke J, Burstein D. Nondestructive imaging of human cartilage glycosaminoglycan concentration by MRI. Magn Reson Med 1999;41: 857e65. 11. Nieminen MT, Rieppo J, Silvennoinen J, To¨yra¨s J, Hakuma¨ki JM, Hyttinen MM, et al. Spatial assessment of articular cartilage proteoglycans with Gd-DTPAenhanced T(1) imaging. Magn Reson Med 2002;48: 640e8. 12. Grampp S, Majumdar S, Jergas M, Newitt D, Lang P, Genant HK. Distal radius: in vivo assessment with quantitative MR imaging, peripheral quantitative CT, and dual X-ray absorptiometry. Radiology 1996;198: 213e8. 13. Kang C, Paley M, Ordidge R, Speller R. In vivo MRI measurements of bone quality in the calcaneus: a comparison with DXA and ultrasound. Osteoporos Int 1999;9:65e74. 14. Arokoski MH, Arokoski JP, Vainio P, Niemitukia LH, Kroger H, Jurvelin JS. Comparison of DXA and MRI

1157

Osteoarthritis and Cartilage Vol. 15, No. 10

15.

16.

17.

18.

19.

20. 21. 22.

23. 24.

25.

26.

27. 28. 29.

30.

31.

methods for interpreting femoral neck bone mineral density. J Clin Densitom 2002;5:289e96. Majumdar S, Newitt D, Jergas M, Gies A, Chiu E, Osman D, et al. Evaluation of technical factors affecting the quantification of trabecular bone structure using magnetic resonance imaging. Bone 1995;17: 417e30. Majumdar S, Genant HK, Grampp S, Newitt DC, Truong VH, Lin JC, et al. Correlation of trabecular bone structure with age, bone mineral density, and osteoporotic status: in vivo studies in the distal radius using high resolution magnetic resonance imaging. J Bone Miner Res 1997;12:111e8. Beuf O, Ghosh S, Newitt DC, Link TM, Steinbach L, Ries M, et al. Magnetic resonance imaging of normal and osteoarthritic trabecular bone structure in the human knee. Arthritis Rheum 2002;46:385e93. Lammentausta E, Kiviranta P, Nissi MJ, Laasanen MS, Kiviranta I, Nieminen MT, et al. T2 relaxation time and delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) of human patellar cartilage at 1.5 T and 9.4 T: relationships with tissue mechanical properties. J Orthop Res 2006;24:366e74. Lammentausta E, Hakulinen MA, Jurvelin J, Nieminen MT. Prediction of mechanical properties of trabecular bone using quantitative MRI. Phys Med Biol 2006;51:6187e98. Nieminen MT, Menezes N, Williams A, Burstein D. T2 of articular cartilage in the presence of Gd-DTPA(2). Magn Reson Med 2004;51:1147e52. Abragam A. The Principles of Nuclear Magnetism. London: Oxford University Press, 1961. Kose K, Matsuda Y, Kurimoto T, Hashimoto S, Yamazaki Y, Haishi T, et al. Development of a compact MRI system for trabecular bone volume fraction measurements. Magn Reson Med 2004;52:440e4. Fernandez-Seara MA, Song HK, Wehrli FW. Trabecular bone volume fraction mapping by low-resolution MRI. Magn Reson Med 2001;46:103e13. To¨yra¨s J, Rieppo J, Nieminen MT, Helminen HJ, Jurvelin JS. Characterization of enzymatically induced degradation of articular cartilage using high frequency ultrasound. Phys Med Biol 1999;44:2723e33. Nieminen MT, To¨yra¨s J, Laasanen MS, Silvennoinen J, Helminen HJ, Jurvelin JS. Prediction of biomechanical properties of articular cartilage with quantitative magnetic resonance imaging. J Biomech 2004;37:321e8. Jurvelin JS, Buschmann MD, Hunziker EB. Optical and mechanical determination of Poisson’s ratio of adult bovine humeral articular cartilage. J Biomech 1997; 30:235e41. Turner CH. Yield behavior of bovine cancellous bone. J Biomech Eng 1989;111:256e60. Turner CH, Burr DB. Basic biomechanical measurements of bone: a tutorial. Bone 1993;14:595e608. Kiviranta I, Jurvelin J, Tammi M, Saamanen AM, Helminen HJ. Microspectrophotometric quantitation of glycosaminoglycans in articular cartilage sections stained with Safranin O. Histochemistry 1985;82: 249e55. Potter K, Kidder LH, Levin IW, Lewis EN, Spencer RG. Imaging of collagen and proteoglycan in cartilage sections using Fourier transform infrared spectral imaging. Arthritis Rheum 2001;44:846e55. Mankin HJ, Dorfman H, Lippiello L, Zarins A. Biochemical and metabolic abnormalities in articular cartilage from osteo-arthritic human hips. II. Correlation of

32.

33.

34. 35.

36. 37. 38.

39.

40. 41. 42.

43.

44.

45.

46.

47.

morphology with biochemical and metabolic data. J Bone Joint Surg Am 1971;53:523e37. Kiraly K, Hyttinen MM, Lapvetela¨inen T, Elo M, Kiviranta I, Dobai J, et al. Specimen preparation and quantification of collagen birefringence in unstained sections of articular cartilage using image analysis and polarizing light microscopy. Histochem J 1997; 29:317e27. Rieppo J, Hallikainen J, Jurvelin JS, Helminen HJ, Hyttinen MM. Novel quantitative polarization microscopic assessment of cartilage and bone birefringence, orientation and anisotropy (Abstract). In: 49th Annual Meeting of the Orthopaedic Research Society. 2003; New Orleans, Louisiana. Zar JH. Biostatical Analysis. 2nd edn. Englewood Cliffs, NJ: Prentice Hall Inc. 1984;313e314. Regatte RR, Akella SV, Lonner JH, Kneeland JB, Reddy R. T1rho relaxation mapping in human osteoarthritis (OA) cartilage: comparison of T1rho with T2. J Magn Reson Imaging 2006;23:547e53. David-Vaudey E, Ghosh S, Ries M, Majumdar S. T2 relaxation time measurements in osteoarthritis. Magn Reson Imaging 2004;22:673e82. Menezes NM, Gray ML, Hartke JR, Burstein D. T2 and T1rho MRI in articular cartilage systems. Magn Reson Med 2004;51:503e9. Wayne JS, Kraft KA, Shields KJ, Yin C, Owen JR, Disler DG. MR imaging of normal and matrix-depleted cartilage: correlation with biomechanical function and biochemical composition. Radiology 2003;228: 493e9. Xia Y, Moody JB, Alhadlaq H. Orientational dependence of T2 relaxation in articular cartilage: a microscopic MRI (microMRI) study. Magn Reson Med 2002;48:460e9. Henkelman RM, Stanisz GJ, Kim JK, Bronskill MJ. Anisotropy of NMR properties of tissues. Magn Reson Med 1994;32:592e601. Xia Y. Relaxation anisotropy in cartilage by NMR microscopy (muMRI) at 14-microm resolution. Magn Reson Med 1998;39:941e9. Regatte RR, Akella SV, Borthakur A, Kneeland JB, Reddy R. Proteoglycan depletion-induced changes in transverse relaxation maps of cartilage: comparison of T2 and T1rho. Acad Radiol 2002;9:1388e94. Watrin-Pinzano A, Ruaud JP, Olivier P, Grossin L, Gonord P, Blum A, et al. Effect of proteoglycan depletion on T2 mapping in rat patellar cartilage. Radiology 2005;234:162e70. Kurkija¨rvi J, Nissi M, Kiviranta I, Jurvelin J, Nieminen MT. Delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) and T2 characteristics of human knee articular cartilage: topographical variation and relationships to mechanical properties. Magn Reson Med 2004;52:41e6. Bobinac D, Spanjol J, Zoricic S, Maric I. Changes in articular cartilage and subchondral bone histomorphometry in osteoarthritic knee joints in humans. Bone 2003;32:284e90. Matsui H, Shimizu M, Tsuji H. Cartilage and subchondral bone interaction in osteoarthrosis of human knee joint: a histological and histomorphometric study. Microsc Res Tech 1997;37:333e42. Toumi H, Higashiyama I, Suzuki D, Kumai T, Bydder G, McGonagle D, et al. Regional variations in human patellar trabecular architecture and the structure of the proximal patellar tendon enthesis. J Anat 2006;208:47e57.