In vivo comparison of delayed gadolinium-enhanced MRI of cartilage and delayed quantitative CT arthrography in imaging of articular cartilage

Osteoarthritis and Cartilage 21 (2013) 434e442 In vivo comparison of delayed gadolinium-enhanced MRI of cartilage and delayed quantitative CT arthrography in imaging of articular cartilage J. Hirvasniemi y a, K.A.M. Kulmala z a, E. Lammentausta x, R. Ojala k, P. Lehenkari {#, A. Kamel x, J.S. Jurvelin z, J. Töyräs zyy, M.T. Nieminen xk, S. Saarakkala yx * y Department of Medical Technology, University of Oulu, Oulu, Finland z Department of Applied Physics, University of Eastern Finland, Kuopio, Finland x Department of Diagnostic Radiology, Oulu University Hospital, Oulu, Finland k Department of Diagnostic Radiology, University of Oulu, Oulu, Finland { Department of Anatomy and Cell Biology, University of Oulu, Oulu, Finland # Department of Surgery, Oulu University Hospital, Oulu, Finland yy Department of Clinical Neurophysiology, Kuopio University Hospital, Kuopio, Finland article info summary Article history: Received 2 July 2012 Accepted 16 December 2012 Keywords: Articular cartilage MRI Contrast-enhanced computed tomography Arthroscopy Osteoarthritis Knee joint Objective: To compare delayed gadolinium-enhanced magnetic resonance imaging (MRI) of cartilage (dgemric) and delayed quantitative computed tomography (CT) arthrography (dqcta) to each other, and their association to arthroscopy. Additionally, the relationship between dgemric with intravenous (dgemric IV ) and intra-articular contrast agent administration (dgemric IA ) was determined. Design: Eleven patients with knee pain were scanned at 3 T MRI and 64-slice CT before arthroscopy. dqcta was performed at 5 and 45 min after intra-articular injection of ioxaglate. Both dgemric IV and dgemric IA were performed at 90 min after gadopentetate injection. dgemric indices and change in relaxation rates (DR 1 ) were separately calculated for dgemric IV and dgemric IA. dgemric and dqcta parameters were calculated for predetermined sites at the knee joint that were International Cartilage Repair Society (ICRS) graded in arthroscopy. Results: dqcta normalized with the contrast agent concentration in synovial fluid (SF) and dgemric IV correlated significantly, whereas dgemric IA correlated with the normalized dqcta only when dgem- RIC IA was also normalized with the contrast agent concentration in SF. Correlation was strongest between normalized dqcta at 45 min and DR 1,IV (r s ¼ 0.72 [95% CI 0.56e0.83], n ¼ 49, P < 0.01) and DR 1,IA normalized with DR 1 in SF (r s ¼ 0.70 [0.53e0.82], n ¼ 52, P < 0.01). Neither dgemric nor dqcta correlated with arthroscopic grading. dgemric IV and non-normalized dgemric IA were not related while DR 1,IV correlated with normalized DR 1,IA (r s ¼ 0.52 [0.28e0.70], n ¼ 50, P < 0.01). Conclusions: This study suggests that dqcta is in best agreement with dgemric IV at 45 min after CT contrast agent injection. dqcta and dgemric were not related to arthroscopy, probably because the remaining cartilage is analysed in dgemric and dqcta, whereas in arthroscopy the absence of cartilage defines the grading. The findings indicate the importance to take into account the contrast agent concentration in SF in dqcta and dgemric IA. Ó 2012 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. * Address correspondence and reprint requests to: S. Saarakkala, Department of Medical Technology, Institute of Biomedicine, University of Oulu, POB 5000, FI-90014 Oulu, Finland. Tel: þ358-400354512. E-mail addresses: jukka.hirvasniemi@oulu.fi (J. Hirvasniemi), katariina.kulmala@ uef.fi (K.A.M. Kulmala), eveliina.lammentausta@oulu.fi (E. Lammentausta), ojala.risto@gmail.com (R. Ojala), petri.lehenkari@oulu.fi (P. Lehenkari), alaaeldin.kamel@ppshp.fi (A. Kamel), jukka.jurvelin@uef.fi (J.S. Jurvelin), juha.toyras@uef.fi (J. Töyräs), miika.nieminen@oulu.fi (M.T. Nieminen), simo.saarakkala@oulu.fi (S. Saarakkala). a Both authors contributed equally. Introduction In osteoarthritis (OA) articular cartilage is progressively degenerated. One of the earliest signs of cartilage degeneration is loss of glycosaminoglycan (GAG) side chains of proteoglycans 1,2, particularly at the superficial layer. Other early degenerative changes in cartilage include the deterioration of the collagen network and an increase in water content 1,3. There are several factors predisposing to OA, including age, obesity, joint injuries, and genetics 2,4. Further, 1063-4584/$ e see front matter Ó 2012 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.joca.2012.12.009

J. Hirvasniemi et al. / Osteoarthritis and Cartilage 21 (2013) 434e442 435 cartilage injuries, e.g., after joint trauma, lead often to the development of OA 2,5. There are different treatment options for focal cartilage lesions 6e8 and thus, early and accurate diagnosis of the cartilage lesions and degenerative changes are important. Clinical diagnosis of OA is based on the physical examination and observation of changes on plain radiographs, occasionally followed by magnetic resonance imaging (MRI) or arthroscopy. Unfortunately, current clinical imaging methods are not sensitive enough to detect cartilage lesions and early OA changes in cartilage. Arthroscopy is still considered as the gold standard for the evaluation of cartilage lesions, although it is based on subjective visualization and palpation, hence, it includes large inter-observer and moderate intra-observer variability 9e11. When aiming at earlier diagnosis of the lesions and early OA changes in cartilage, noninvasive determination of structural and compositional changes of cartilage tissue is an advantage. One approach to probe the composition of articular cartilage is to use negatively charged contrast agent to enhance MR or computed tomography (CT) imaging. These methods are based on assumption that negatively charged contrast agent distributes into cartilage in an inverse relation to the fixed charge density in cartilage associated with GAG content of the cartilage 12,13. Therefore, higher concentration of anionic contrast agent diffuses into degraded cartilage than into intact cartilage. Diffusion and distribution of contrast agent are, however, influenced also by other factors in cartilage, e.g., collagen and water content 14e17. With regard to MRI, a technique called delayed gadoliniumenhanced MRI of cartilage (dgemric) is based on the aforementioned properties and it has been proposed for quantitative estimation of the GAG concentration in cartilage 12,13,18. The dgemric method has been applied both in vitro 18e20 and in vivo 13,21,22. While the specificity of dgemric to GAG has been recently questioned 14,17,23, it is reported to sensitively detect degenerative changes in cartilage 21. In the dgemric method, the contrast agent (gadopentetate, charge 2) can be administered either intravenously (dgemric IV ) or intra-articularly (dgemric IA ). Intravenously injected contrast agent may enter the cartilage both through the cartilage surface and from the subchondral bone whereas intra-articularly injected contrast agent can diffuse only through the surface 13. Although diffusion of the contrast agent from the subchondral bone was slow or negligible in recent in vitro and in vivo studies 14,17,23, the transportation of the contrast agent into the cartilage in the intravenous and intra-articular dgemric methods may be different. At the moment, there are no in vivo studies comparing the intravenous and intra-articular dgemric methods in a knee joint of a same patient. Contrast-enhanced CT, an analogous X-ray technique to dgemric, also employs contrast agent (e.g., anionic ioxaglate, charge 1). There are a range of in vitro studies in which the contrast-enhanced CT technique has shown its potential in assessment of GAG content 24e28 and biomechanical properties 24,29 of cartilage, as well as in detection of cartilage injuries 30,31. Contrast-enhanced CT, referred to as delayed quantitative CT arthrography (dqcta) in the present study, has recently been tested in vivo 32, but it has not been thoroughly validated in clinical settings. Although the dgemric and contrast-enhanced CT techniques were initially designed to probe GAG content of the cartilage as well as degenerative stage of cartilage 12,13,25,33e36, they have not been systematically compared in vivo for the same patients. In the present study, both dgemric and dqcta were conducted in vivo for patients referred to a knee arthroscopy because of knee pain symptoms. The hypotheses of the study were: (1) a strong linear correlation between dgemric and dqcta parameters should be found, (2) both dgemric and dqcta parameters should be related to arthroscopic grading of cartilage, and (3) dgemric IV should be significantly related to dgemric IA. Methods Study subjects Eleven consecutive patients (eight females and three males) referred to an arthroscopic surgery of the knee because of persistent knee pain symptoms were enrolled in the present study (Table I). One patient declined arthroscopy but completed all imaging studies and one patient was excluded from the analysis due to irregular distribution of contrast agent in joint. Before arthroscopy, MRI (three imaging sessions) and CT (two imaging sessions) examinations were performed as described in Fig. 1. Informed consent was obtained from all patients. The study was approved by the Ethical Committee of the Northern Ostrobothnia Hospital District, Oulu, Finland (No. 33/2010). MRI For MRI, each patient was scanned three times on a 3 T scanner (Siemens Skyra, Siemens Healthcare, Erlangen, Germany) with a dedicated 15-channel transmit/receive knee coil (Quality Electrodynamics (QED), MayField Village, OH, USA). For anatomical imaging, double echo steady state (DESS) sequence with water excitation (repetition time (TR)/time to echo (TE) ¼ 14.1/5 ms, field of view (FOV) ¼ 150*150 mm 2, matrix ¼ 256*256, slice thickness ¼ 0.6 mm) was performed in first imaging session whereas T 1 relaxation times were measured in all three sessions. Prior to contrast agent administration, single-slice T 1 mapping was performed at the centre of medial and lateral condyles using an inversion recovery fast spin echo (IR-FSE) sequence (TR/TE/ inversion time (TI) ¼ 4060/8.6/50, 100, 200, 400, 800, 1600, 3200, and 3900 ms; FOV ¼ 120*120 mm 2 ; matrix ¼ 256*256; slice thickness ¼ 3 mm). Subsequently, 0.2 mm/kg (double dose) of gadopentetate (Gd-DTPA 2, MagnevistÔ) was injected intravenously, followed by active flexion-extension exercises of the knee for 5 min and walking for 5 min. T 1 measurements were repeated at 90 min after intravenous administration of Gd-DTPA 2 using the same imaging parameters 21,33. Two weeks after the previous imaging session, dgemric IA was performed after a 20 ml dose of an ioxaglate e Gd-DTPA 2 contrast agent mixture (105 mm HexabrixÔ 320, Guerbet, Roissy, France and 2.5 mm MagnevistÔ, Bayer HealthCare Pharmaceuticals, Berlin, Germany; diluted in 0.9% saline; Table I Description of the patients and their preliminary diagnosis according to international classification of diseases (ICD)-10 codes Patient Gender Age (years) Height (cm) Weight (kg) Body mass index (BMI) (kg/m 2 ) Preliminary diagnosis (ICD-10) 1 Male 66 170 71 24.6 M23.2 2 Male 59 176 101 32.6 M23.2 3 Female 55 165 75 27.5 M23.2 4 Female 63 167 73 26.2 M23.2 5 Female 61 167 70 25.1 S83.2 6 Female 40 163 73 27.5 M17.1 7 Female 50 168 90 31.9 M23.2 8 Female 68 152 67 29.0 S83.2 9 Female 55 170 98 33.9 M23.2 10 Female 58 164 55 20.4 M23.2 M17.1 ¼ other primary arthrosis of the knee. M23.2 ¼ derangement of meniscus due to old tear or injury. S83.2 ¼ current tear of meniscus.

436 J. Hirvasniemi et al. / Osteoarthritis and Cartilage 21 (2013) 434e442 osmolality: 370 mosmol/kg of water) was injected intra-articularly (see CT section below). The post-contrast T 1 measurements were performed at 90 min after intra-articular injection using the aforementioned imaging protocol. dgemric analysis Pre-contrast MRI Intravenous injection of Gd-DTPA 2-90 min Post-contrast MRI at 90 minutes after the intravenous contrast agent injection (dgemric IV ) 2weeks Intra-articular injection of the combination of Gd-DTPA 2- and ioxaglate 5min CT scan at 5 minutes after the intra-articular contrast agent injection 40 min CT scan at 45 minutes after the intra-articular contrast agent injection 45 min Post-contrast MRI at 90 minutes after the intra-articular contrast agent injection (dgemric IA ) 2-12 weeks Arthroscopy of the knee Fig. 1. The flow diagram of the study. An in-house MATLAB application (v.7.9.0; MathWorks inc., Natick, MA, USA) was used to generate T 1 maps with a pixel-by-pixel three-parameter fit routine. Articular cartilage was segmented manually for the quantitative T 1 analysis. Six region of interests (ROI) corresponding to those in arthroscopic evaluation were extracted from the segmented cartilage, including medial trochlear groove (MTG) and lateral trochlear groove (LTG) and condyles of tibia and femur. In addition, a ROI from synovial fluid (SF) was segmented. First voxel near at the surface of the cartilage as well as at the cartilageebone interface was excluded from the analyses to avoid partial volume effect. The mean T 1 relaxation time of each cartilage ROI, i.e., the dgemric index, was separately calculated for dgemric IV and dgemric IA (T 1Gd,IV and T 1Gd,IA, respectively). Additionally, the change in relaxation rate was calculated for cartilage and SF (DR 1,IV, DR 1,IA, and DR 1,SF ) as follows 33 : DR 1 ¼ 1=T 1Gd 1=T 1;0 ; (1) where T 1Gd and T 1,0 are relaxation time values with and without Gd-DTPA 2, respectively. Furthermore, since the contrast agent is diluted in the SF after the intra-articular injection, the DR 1,IA was normalized by the DR 1,SF (¼DR 1,IA /DR 1,SF ) analogously to the normalized dqcta parameters. As patella (PAT) was not properly visualized in the dgemric slices, it was excluded from the analyses. Furthermore, whole cartilage surface was not visible in all dgemric slices and thus, the eventual sample sizes for T 1Gd,IV and T 1Gd,IA were 53 and 59, respectively. Sample sizes for DR 1,IV, DR 1,IA, and DR 1,IA /DR 1,SF were 50, 54, and 54, respectively. CT Before CT scan, the ioxaglate - Gd-DTPA 2 contrast agent mixture was injected intra-articularly (see MRI). The contrast agent mixture was prepared in hospital pharmacy under sterile conditions. Gadopentetate concentration was similar to those typically used in MR-arthrography 37,38. Ioxaglate concentration was lower than used in CT arthrography 39, since hyperosmolaric contrast agent may cause transient softening of articular cartilage and negatively affect cartilage if the joint is excessively loaded after injection 40. After injection, the patient performed active flexionextension of the knee for 5 min to enable a smooth distribution of the contrast agent into joint surfaces. After this exercise period, the knee joint of the patient was scanned using a clinical 64-slice CT-scanner (DiscoveryÔ PET/CT 690, GE Medical Systems, Waukesha, WI, USA) with a tube voltage of 100 kv and current of 160 ma. Focal spot size was 0.7 mm and pitch was 0.53. After the first imaging session, the patient was rescanned at 45 min after the contrast agent injection. The patients avoided walking between CT imaging sessions. CT scan at 5 min after injection was selected to correspond with conventional CT arthrography, enabling also the segmentation of cartilage. It has been suggested that the contrast agent concentration in cartilage is near to its maximum between 30 and 60 min after injection 32, and therefore a 45 min time point for the second CT scan was selected. dqcta analysis Analyze software (Analyze 10.0, AnalyzeDirect, Inc., Overland Park, KS, USA) was used in image processing. Since the original resolution varied between images (232e447 mm), they were resampled to the same isotropic voxel size (312 312 312 mm 3 ) using linear interpolation. The segmentation of the cartilage was conducted using a region growing method. Dimensions of the 3-D ROI were variable, but they did not exceed the limits that were set to be one-third of the studied joint surface width. The ROI was located in the centre of the studied joint surface. Seven cartilage ROIs were extracted from the same locations as in the dgemric analyses and, in addition, from the PAT. Furthermore, a SF ROI was selected from within the largest and clearest SF volume in both CT and MRI. Doubtful voxels were excluded to avoid partial volume effect. The mean X-ray attenuation values of the cartilage (C) and SF ROIs at 5 and 45 min after contrast agent injection (C 5,C 45,SF 5, and SF 45 ) were calculated to evaluate the contrast agent concentrations. The contrast agent becomes diluted in SF and the volume of the SF varies among patients. Therefore, to eliminate any variation in the equilibrating contrast agent concentration, the cartilage parameters at each time point were normalized by the contrast agent concentration in the SF at the same time point, i.e., C 5 /SF 5 and C 45 /SF 45. Since the seed point for the region growing could not be defined for all cases due to absence of cartilage, the eventual sample size in dqcta analyses was 67. Arthroscopy After imaging studies, arthroscopy was conducted for the patients (n ¼ 9) within 2e12 (mean ¼ 5.4, standard deviation

J. Hirvasniemi et al. / Osteoarthritis and Cartilage 21 (2013) 434e442 437 (SD) ¼ 2.8) weeks. During arthroscopy, the predetermined sites at knee joint (medial condyle of tibia (TMC), medial condyle of femur (FMC), lateral condyle of tibia (TLC), lateral condyle of femur (FLC), MTG, LTG, and PAT) were classified according to the International Cartilage Repair Society (ICRS) grading system 41 by an experienced orthopaedic surgeon (PL). In the ICRS grading scale, intact cartilage is graded as 0, slightly softened indentation stiffness and/or superficial fissures and cracks as 1, lesions extending less than 50% of cartilage depth as 2, cartilage lesions extending 50% or more of cartilage depth but not into the subchondral bone as 3, and cartilage lesions extending into the subchondral bone as 4. ICRS grading was missing from four sites and thus, the total number of ICRS graded sites was 59. Statistical analysis Statistical analyses were performed using SPSS 19 software (SPSS Inc., Chicago, IL, USA). The level of statistical significance was set to P < 0.05. Normal distribution of the parameters was tested using KolmogoroveSmirnov test and based on the normality of the parameters either Pearson (r) or Spearman (r s ) correlation analysis (together with 95% confidence intervals [CI]) 42 was applied. Differences between the correlation coefficients were statistically compared by using Fisher s transform 43. KruskaleWallis test was used for group comparisons. Results The mean T 1 relaxation time values in the cartilage were 59% (range among the different anatomical sites: 46e68%) and 45% (34e50%) lower after intravenous and intra-articular contrast agent injection than without contrast agent, respectively, whereas the SF relaxation time values were 90% and 67% lower, respectively (Table II). DR 1 values in the cartilage were 76% (44e124%) higher after intravenous injection than after intra-articular injection (Table II). The mean X-ray attenuation in the cartilage was 27% (22e38%) higher whereas the attenuation in the SF was 62% lower at 45 min than at 5 min after contrast agent injection (Table II). The mean C 45 /SF 45 value was 222% (209e254%) higher than the mean C 5 /SF 5 value (Table II). Comparison of dgemric and dqcta Representative MR and CT images, T 1 relaxation time map overlaid on top of an MR image, and illustrative dqcta map of C 45 / SF 45 overlaid on top of a CT image of a patient with ICRS grade 2 cartilage lesion are presented in Fig. 2. In quantitative analyses, dgemric IV correlated strongest with normalized dqcta parameters (Table III, Fig. 3), while dgemric IA correlated strongest with dqcta at 45 min after the both parameters were normalized with contrast agent concentration in SF (Table III). When studying relation between dgemric and dqcta in intact and slightly degraded cartilage (ICRS ¼ 0e1), DR 1,IV and DR 1,IA / DR 1,SF correlated with C 45 /SF 45 (r ¼ 0.65 [0.36e0.82], n ¼ 28, P < 0.01 and r s ¼ 0.81 [0.63e0.91], n ¼ 29, P < 0.01, respectively). These correlations were not significantly different from those in the whole population (ICRS ¼ 0e3). Association of arthroscopy to dgemric and dqcta In the arthroscopy, ICRS classification for the predetermined cartilage surfaces varied from normal (grade 0, n ¼ 19) and nearly normal (grade 1, n ¼ 19) to abnormal (grade 2, n ¼ 19) and severely abnormal (grade 3, n ¼ 2). dqcta showed a trend towards increasing values with severity of cartilage damage seen in arthroscopy. However, dqcta and dgemric parameters were not statistically significantly different in different ICRS grade groups (Fig. 4). Association of dgemric IV and dgemric IA There was no association between dgemric IV and dgemric IA when correlating either T 1 relaxation time values (r ¼ 0.12 [ 0.38e0.16], n ¼ 53, P ¼ 0.39) or DR 1 values (r s ¼ 0.01 [ 0.29e 0.27], n ¼ 50, P ¼ 0.95). When DR 1,IA was normalized by the DR 1,SF,a significant correlation to DR 1,IV was established (r s ¼ 0.52 [0.28e 0.70], n ¼ 50, P < 0.01). Discussion In the present study, we hypothesized that there is a strong relationship between the dgemric and dqcta techniques due to the similar methodological approach used with these two different modalities. This hypothesis was confirmed, yet, the relation was dependent on whether the parameters were normalized with contrast agent concentration in SF or not. Second, we hypothesized that dgemric and dqcta would be related to arthroscopic ICRS grading of cartilage. However, this association was not observed. Third, we hypothesized that there is a relationship between dgemric IV and dgemric IA. The relationship between these two dgemric methods was significant only when the DR 1,IA was normalized by the DR 1 in SF. The correlation between dgemric and dqcta was expected, since both of these methods are based on the use of the anionic contrast agent and proposed for the quantification of the cartilage Table II The mean (SD) X-ray attenuation values (Hounsfield Units) and normalized dqcta parameters at 5 and 45 min after contrast agent injection, and the mean (SD) T 1 relaxation time (ms) and DR 1 (1/s) values in SF and in cartilage. The mean X-ray attenuation in the cartilage was higher at 45 min after injection than at 5 min after injection. The mean attenuation was lower in the SF at 45 min after injection that at 5 min after injection. The mean T 1 relaxation time values in the cartilage and in the SF were lower after intravenous and intra-articular contrast agent administration than without contrast agent CT at 5 min CT at 45 min C 5 /SF 5 C 45 /SF 45 T 1,0 T 1Gd,IV T 1Gd,IA DR 1,IV DR 1,IA SF 1168 294 439 80 e e 3427 367 345 30* 1122 385 e 0.71 0.38 PAT 184 38 228 53 0.17 0.05 0.53 0.15 e e e e e LTG 154 18 194 28 0.14 0.04 0.46 0.07 1337 121 536 63 742 118 1.14 0.28 0.62 0.18 MTG 164 32 211 46 0.16 0.05 0.49 0.12 1441 250 550 98 748 115 1.02 0.25 0.70 0.20 FLC 136 16 187 26 0.12 0.03 0.44 0.09 1402 303 535 51 741 146 1.14 0.28 0.65 0.17 FMC 198 47 253 55 0.19 0.08 0.59 0.16 1493 174 484 122 752 149 1.54 0.61 0.69 0.19 TLC 147 49 183 66 0.14 0.06 0.43 0.19 1077 122 578 102 711 67 0.85 0.36 0.48 0.08 TMC 213 69 260 74 0.19 0.06 0.60 0.18 1041 203 484 93 584 94 1.27 0.57 0.79 0.24 C x ¼ cartilage, x ¼ time in minutes from contrast agent injection, T 1,0 ¼ T 1 relaxation time without contrast agent, IV ¼ intravenously administered contrast agent, IA ¼ intraarticularly administered contrast agent. * T 1 relaxation time in SF could not be defined for two patients.

438 J. Hirvasniemi et al. / Osteoarthritis and Cartilage 21 (2013) 434e442 Fig. 2. Sagittal MR (AeC) and CT (DeF) images from the corresponding location in a patient with ICRS grade 2 cartilage lesion (arrow). According to visual evaluation, CT conducted at 5 min after the contrast agent injection had the best diagnostic quality for evaluation of cartilage lesions. (A) Anatomical DESS image (TE/TR ¼ 5/14.1 ms) prior to contrast agent administration. (B) IR-FSE image (TI/TE/TR ¼ 200/8.6/4060 ms) prior to contrast agent administration (note: cartilage appears thinner as compared to DESS due to longer TR). (C) T 1 relaxation time map of cartilage overlaid on top of a T 1 -weighted MR image after intravenous contrast agent administration. (D) CT at 5 min and (E) CT at 45 min after contrast agent injection. (F) Illustrative dqcta map of normalized X-ray attenuation in cartilage at 45 min after contrast agent injection (C 45 /SF 45 ) overlaid on top of an anatomical CT image. Contrast of the images has been adjusted to enhance visibility of the lesion. GAG content 12,13,18,24e27. However, the correlations varied significantly depending on which parameters were used to quantify the contrast agent concentration in cartilage. The strongest correlations between dgemric IV and dqcta were obtained when dqcta parameters were normalized with the contrast agent concentration in SF. Further, strongest correlation between dgemric IA and normalized dqcta was established when dgemric IA was normalized by the DR 1 in SF. These findings underline the importance to consider the contrast agent concentration in SF in dgem- RIC IA and dqcta analyses, particularly in patient populations with expected differences in SF volume. dqcta and dgemric were similarly related even after excluding severe degenerative cases (ICRS ¼ 2e3) from the analyses. This result suggests that these two techniques share similar features also when only intact or slightly degenerated cartilage is investigated. Pharmacokinetically, dgem- RIC at 90 min is closest to the dqcta at 45 min, yet the authors acknowledge the considerable difference in time between these two measurements. However, the selection of imaging time points was based on the literature 21,32,33. It is likely that at 5 min after intra-articular injection, the contrast agent was poorly diffused into cartilage and thus, the correlations with dgemric parameters were lower. Differences in pharmacokinetics between intra-articular and intravenous administration and between the applied contrast agents may explain some of the variation in correlations. It has been proposed that the transportation of the contrast agent into the cartilage in intravenous and intra-articular injection may be different 13, although the diffusion from the subchondral bone was negligible in recent studies 14,17,23. Furthermore, it was previously shown that the penetration of gadopentetate and ioxaglate into cartilage is different due to their different molecular masses and charges (gadopentetate: 548 g/mol, 2; ioxaglate: 1269 g/ mol, 1) 14,44. The mean attenuation in the SF decreased by 62% between the 5 and 45 min time points parallel with the results in a previous study with two patients CT scanned before and at 30, 60, and 120 min after intra-articular injection of 50% ioxaglate solution 32. It should be noted that a high correlation (r s ¼ 0.91, P < 0.01) was observed between DR 1,SF and SF 45, indicating that the initial equilibrating conditions for dqcta and dgemric IA were similar. Neither dqcta nor dgemric parameters were significantly different between different ICRS grades. Thus the hypothesis that both quantitative dqcta and dgemric are related to arthroscopic ICRS grading, could not be confirmed. This finding is in agreement with two previous studies in which the dgemric parameters did not correlate with the arthroscopic findings 35,45. Owman et al. (2008) calculated the dgemric index for the TLC, FLC and TMC, FMC 35 whereas Nojiri et al. (2006) calculated it only for the lesion in the PAT 45. However, those and the present results contradict with Table III Spearman rank correlation coefficients (95% CI) between dgemric and dqcta parameters. Strongest correlations were observed when dqcta parameters were normalized with contrast agent concentration in SF and DR 1,IA was normalized with change in relaxation rate in SF C 5 C 45 C 5 /SF 5 C 45 /SF 45 DR 1,IV 0.28 ( 0.01e0.52) 0.39 (0.12e0.60)** 0.42 (0.16e0.63)** 0.72 (0.56e0.83)** DR 1,IA 0.42 (0.16e0.62)** 0.42 (0.17e0.62)** 0.27 ( 0.01e0.50) 0.06 ( 0.22e0.32) DR 1,IA /DR 1,SF 0.10 ( 0.18e0.36) 0.16 ( 0.12e0.42) 0.13 ( 0.15e0.39) 0.70 (0.53e0.82)** T 1Gd,IV 0.31 ( 0.54 to 0.04)* 0.42 ( 0.62 to 0.16)** 0.48 ( 0.66 to 0.23)** 0.68 ( 0.80 to 0.50)** T 1Gd,IA 0.43 ( 0.62 to 0.18)** 0.42 ( 0.61 to 0.18)** 0.26 ( 0.49e0.01) 0.03 ( 0.24e0.29) *P < 0.05, **P < 0.01. C x ¼ cartilage, x ¼ time in minutes from contrast agent injection. IV ¼ intravenously administered contrast agent. IA ¼ intra-articularly administered contrast agent.

J. Hirvasniemi et al. / Osteoarthritis and Cartilage 21 (2013) 434e442 439 Fig. 3. Statistically significant positive correlation between normalized mean attenuation in cartilage at 45 min after contrast agent injection (C 45 /SF 45 ) and DR 1,IV was observed. Linear fit is for illustrative purposes. another study that reported significantly different R 1 and DR 1 values between the OA and reference compartments 33. This may be due to the fact that Tiderius et al. (2003) limited their analysis to the load bearing FLC and FMC, representing a more homogenous selection of cartilage. All those three studies were conducted at 1.5 T MRI at 90e120 min after intravenous injection of Gd-DTPA 2 (0.1e0.3 mm/kg) 33,35,45. The present study consisted of ROIs at various locations, and cartilage properties including the dgemric index have previously been shown to vary across different cartilage surfaces 46,47. Further, the remaining cartilage is analysed in the quantitative dgemric and dqcta analyses, whereas in arthroscopy the absence of cartilage defines the classification. Hence, when there is an observed cartilage lesion in the arthroscopy, the quality of the underlying cartilage is actually evaluated with the dgemric and dqcta techniques. However, the remaining cartilage layer just under the lesion should likely have also at least some degenerative changes. Furthermore, the cartilage analysed in the dgemric or dqcta may not cover the whole lesion evaluated in arthroscopy. Theoretically, the dgemric and contrast-enhanced CT methods may detect changes in the cartilage earlier than they are manifested in the arthroscopy. This may allow earlier intervention in disease process with an appropriate approach and these imaging methods may also help to develop new disease-modifying drugs for OA by enabling more accurate and minimally invasive (compared to arthroscopy) follow-up of cartilage changes. However, to confirm those hypotheses, further studies are required. When comparing dgemric IV and dgemric IA parameters, we found a moderate correlation only between DR 1,IA /DR 1,SF and DR 1,IV. Correlation between the dgemric indices or between the nonnormalized DR 1,IA and DR 1,IV was not significant. Several reasons can be suggested to explain this finding. First, the intra-articular contrast agent concentration will be diluted by the SF, and the SF volume may vary among patients, as discussed above. Second, the volume of contrast agent for intra-articular administration was constant for all patients while the volume for intravenous administration depended on the weight of the patient. Third, as discussed earlier, contrast agent diffusion into cartilage may not be similar in intravenous and intra-articular injection, although the diffusion from the subchondral bone was found to be negligible in a recent in vivo study with healthy volunteers scanned at 1.5 T MRI before and at 12e240 min after intravenous injection of Gd-DTPA 2 (0.3 mm/kg) 23. Fourth, the time delays between the post-contrast imaging, yet similar for intra-articular and intravenous administration in this study, may further influence the accumulation of contrast agent into cartilage 13. On average, the relaxation time values were 26% lower and DR 1 values were 76% higher in the intravenous injection than in the intra-articular injection, indicating that contrast agent concentration in the cartilage was higher after intravenous injection. The optimal delay for intra-articular imaging has not been sufficiently investigated and requires further investigation. In the knee, the typical time delay between intravenous injection and imaging has been 90e180 min 21,33,48.In Fig. 4. Comparison of dqcta (A: C 5 /SF 5 and B: C 45 /SF 45 ) and dgemric (C: DR 1,IV and D: DR 1,IA /DR 1,SF ) parameters between different ICRS grade groups. Differences between the different ICRS grade groups were not significant. ICRS grades 2 and 3 were merged into one group due to low number of lesions with grade 3.

440 J. Hirvasniemi et al. / Osteoarthritis and Cartilage 21 (2013) 434e442 the study of Bashir et al. (1997) intra-articular contrast agent was combined with epinephrine to slow down the extraction of the contrast agent and imaging was performed 2.5e7 h after intraarticular injection of the contrast agent. In the present study, epinephrine was not used. This study contains limitations that need to be addressed. First, the time delay between contrast agent injection and imaging was not identical in dgemric IA and dqcta, since we chose to inject both MR and CT contrast agents at the same time for ethical reasons. Second, intra-articular injection is more challenging to conduct than intravenous injection, which was also observed in the present study. Due to anatomical variations between patients, contrast agent did not distribute uniformly into joint surfaces in one intra-articular injection. This injection was given to suprapatellar recess of the knee, which is the most common injection site to knee joint 49. There is generally a wide communication within the knee joint, but rarely, this pouch can remain isolated and independent of the synovial cavity. The remaining injections were given straight to the knee joint. This is recommended to keep the intra-articular concentration of the contrast agent stable. Third, segmentation procedures were different in dgemric and dqcta, however, the agreement of segmentation was confirmed visually. Unlike for MRI, automated segmentation in dqcta was performed since this can be more easily conducted due to the higher contrast between cartilage and adjacent tissues. As compared to CT, image acquisition takes longer time in dgemric and imaging single slices per condyle in dgemric made the imaging time reasonable. Fourth, the dqcta and dgemric analyses were not directly co-registered, and consequently the cartilages between different modalities may not fully correspond to each other. Although the analyses were not directly co-registered, they were still very closely matched by visual evaluation and by selecting similar slice orientation in CT as in dgemric. Finally, as discussed earlier, the dgemric index for cartilage varies across and within joint surfaces 46,47. Perceiving this fact, we did not calculate site-specific correlations between dgemric or dqcta parameters and ICRS grading due to limited sample size. This confounds the interpretation of the present results. This study suggests that the dqcta is in best agreement with dgemric IV at 45 min after ioxaglate injection. However, if judged only with visual evaluation by the experienced radiologist (RO), the CT conducted at 5 min after the injection had the best diagnostic quality for evaluation of cartilage lesions. Thus, two separate scans might offer the optimal result for both qualitative and quantitative evaluation of cartilage. However, both logistics and radiation exposure may hinder the application of two separate scans, although low radiation dose procedures for the imaging of the knee joint have been introduced 50. Thus, further studies are still needed to optimize the imaging delay for dqcta. In conclusion, normalized dqcta at 45 min time point was related to dgemric IV and to normalized dgemric IA parameters. dqcta and dgemric were not related to arthroscopic grading, probably because the remaining cartilage is analysed in the dgemric and dqcta analyses, whereas in arthroscopy the absence of cartilage defines the classification. dgemric IV and dgemric IA were related after change in relaxation rate in SF was taken into account in dgemric IA analyses. The findings of the study indicate the importance to take into account the contrast agent concentration in SF in dqcta and dgemric IA. Author contributions Jukka Hirvasniemi: conception and design, analysis and interpretation of the data, drafting and final approval of the article. Katariina A.M. Kulmala: conception and design, analysis and interpretation of the data, drafting and final approval of the article. Eveliina Lammentausta: conception and design, interpretation of the data, critical revision for important intellectual content and final approval of the article. Risto Ojala: conception and design, interpretation of the data, critical revision for important intellectual content and final approval of the article. Petri Lehenkari: conception and design, collection of the data (arthroscopic grading), critical revision for important intellectual content and final approval of the article. Alaaeldin Kamel: collection of the data, critical revision for important intellectual content and final approval of the article. Jukka S. Jurvelin: conception and design, interpretation of the data, critical revision for important intellectual content and final approval of the article. Juha Töyräs: conception and design, interpretation of the data, critical revision for important intellectual content and final approval of the article. Miika T. Nieminen: conception and design, interpretation of the data, critical revision for important intellectual content and final approval of the article. Simo Saarakkala: conception and design; interpretation of the data; drafting, critical revision for important intellectual content, and final approval of the article. Role of the funding source This study was financially supported by the Sigrid Jusélius Foundation, Päivikki and Sakari Sohlberg Foundation, Academy of Finland (projects 127198 and 128603), and strategic funding of the University of Oulu (project 24001200) and University of Eastern Finland. The funding sources had no role in the study design, data collection or analysis, interpretation of data, writing of the manuscript, or in the decision to submit the manuscript for publication. Conflict of interest The authors report no conflicts of interests. Acknowledgements Esa Liukkonen, PhD, is acknowledged for his assistance in study logistics. References 1. Armstrong CG, Mow VC. Variations in the intrinsic mechanical properties of human articular cartilage with age, degeneration, and water content. J Bone Joint Surg Am 1982;64:88e94. 2. Buckwalter JA, Mankin HJ. Articular cartilage: degeneration and osteoarthritis, repair, regeneration, and transplantation. Instr Course Lect 1998;47:487e504. 3. Maroudas A, Venn M. Chemical composition and swelling of normal and osteoarthrotic femoral head cartilage II. Swelling. Ann Rheum Dis 1977;36:399e406. 4. Arden N, Nevitt MC. Osteoarthritis: epidemiology. Baillieres Best Pract Res Clin Rheumatol 2006;20:3e25. 5. Lotz MK, Kraus VB. New developments in osteoarthritis. Posttraumatic osteoarthritis: pathogenesis and pharmacological treatment options. Arthritis Res Ther 2010;12:211. 6. Bedi A, Feeley BT, Williams 3rd RJ. Management of articular cartilage defects of the knee. J Bone Joint Surg Am 2010;92: 994e1009. 7. Brittberg M, Lindahl A, Nilsson A, Ohlsson C, Isaksson O, Peterson L. Treatment of deep cartilage defects in the knee

J. Hirvasniemi et al. / Osteoarthritis and Cartilage 21 (2013) 434e442 441 with autologous chondrocyte transplantation. N Engl J Med 1994;331:889e95. 8. Hangody L, Fules P. Autologous osteochondral mosaicplasty for the treatment of full-thickness defects of weight-bearing joints: ten years of experimental and clinical experience. J Bone Joint Surg Am 2003;85-A:25e32. 9. Spahn G, Klinger HM, Baums M, Hoffmann M, Plettenberg H, Kroker A, et al. Near-infrared spectroscopy for arthroscopic evaluation of cartilage lesions: results of a blinded, prospective, interobserver study. Am J Sports Med 2010;38:2516e 21. 10. Spahn G, Klinger HM, Hofmann GO. How valid is the arthroscopic diagnosis of cartilage lesions? Results of an opinion survey among highly experienced arthroscopic surgeons. Arch Orthop Trauma Surg 2009;129:1117e21. 11. Brismar BH, Wredmark T, Movin T, Leandersson J, Svensson O. Observer reliability in the arthroscopic classification of osteoarthritis of the knee. J Bone Joint Surg Br 2002;84:42e7. 12. Bashir A, Gray ML, Burstein D. Gd-DTPA2 as a measure of cartilage degradation. Magn Reson Med 1996;36:665e73. 13. Bashir A, Gray ML, Boutin RD, Burstein D. Glycosaminoglycan in articular cartilage: in vivo assessment with delayed Gd(DTPA)(2 )-enhanced MR imaging. Radiology 1997;205: 551e8. 14. Silvast TS, Kokkonen HT, Jurvelin JS, Quinn TM, Nieminen MT, Toyras J. Diffusion and near-equilibrium distribution of MRI and CT contrast agents in articular cartilage. Phys Med Biol 2009;54:6823e36. 15. Evans RC, Quinn TM. Solute diffusivity correlates with mechanical properties and matrix density of compressed articular cartilage. Arch Biochem Biophys 2005;442:1e10. 16. Piscaer TM, Waarsing JH, Kops N, Pavljasevic P, Verhaar JA, van Osch GJ, et al. In vivo imaging of cartilage degeneration using microct-arthrography. Osteoarthritis Cartilage 2008;16: 1011e7. 17. Salo EN, Nissi MJ, Kulmala KA, Tiitu V, Toyras J, Nieminen MT. Diffusion of Gd-DTPA2 into articular cartilage. Osteoarthritis Cartilage 2012;20:117e26. 18. Bashir A, Gray ML, Hartke J, Burstein D. Nondestructive imaging of human cartilage glycosaminoglycan concentration by MRI. Magn Reson Med 1999;41:857e65. 19. Nieminen MT, Rieppo J, Silvennoinen J, Toyras J, Hakumaki JM, Hyttinen MM, et al. Spatial assessment of articular cartilage proteoglycans with Gd-DTPA enhanced T1 imaging. Magn Reson Med 2002;48:640e8. 20. Nissi MJ, Toyras J, Laasanen MS, Rieppo J, Saarakkala S, Lappalainen R, et al. Proteoglycan and collagen sensitive MRI evaluation of normal and degenerated articular cartilage. J Orthop Res 2004;22:557e64. 21. Dahlberg LE, Lammentausta E, Tiderius CJ, Nieminen MT. In vivo monitoring of joint cartilage e lessons to be learned by delayed gadolinium enhanced magnetic resonance imaging of cartilage. Eur Musculoskelet Rev 2012;7:58e62. 22. Williams A, Gillis A, McKenzie C, Po B, Sharma L, Micheli L, et al. Glycosaminoglycan distribution in cartilage as determined by delayed gadolinium-enhanced MRI of cartilage (dgemric): potential clinical applications. AJR Am J Roentgenol 2004;182:167e72. 23. Hawezi ZK, Lammentausta E, Svensson J, Dahlberg LE, Tiderius CJ. In vivo transport of Gd-DTPA(2 ) in human knee cartilage assessed by depth-wise dgemric analysis. J Magn Reson Imaging 2011;34:1352e8. 24. Bansal PN, Joshi NS, Entezari V, Grinstaff MW, Snyder BD. Contrast enhanced computed tomography can predict the glycosaminoglycan content and biomechanical properties of articular cartilage. Osteoarthritis Cartilage 2010;18:184e91. 25. Cockman MD, Blanton CA, Chmielewski PA, Dong L, Dufresne TE, Hookfin EB, et al. Quantitative imaging of proteoglycan in cartilage using a gadolinium probe and microct. Osteoarthritis Cartilage 2006;14:210e4. 26. Siebelt M, van Tiel J, Waarsing JH, Piscaer TM, van Straten M, Booij R, et al. Clinically applied CT arthrography to measure the sulphated glycosaminoglycan content of cartilage. Osteoarthritis Cartilage 2011;19:1183e9. 27. Xie L, Lin AS, Guldberg RE, Levenston ME. Nondestructive assessment of sgag content and distribution in normal and degraded rat articular cartilage via EPIC-microCT. Osteoarthritis Cartilage 2010;18:65e72. 28. Kokkonen HT, Makela J, Kulmala KA, Rieppo L, Jurvelin JS, Tiitu V, et al. Computed tomography detects changes in contrast agent diffusion after collagen cross-linking typical to natural aging of articular cartilage. Osteoarthritis Cartilage 2011;19:1190e8. 29. Aula AS, Jurvelin JS, Toyras J. Simultaneous computed tomography of articular cartilage and subchondral bone. Osteoarthritis Cartilage 2009;17:1583e8. 30. Kokkonen HT, Jurvelin JS, Tiitu V, Toyras J. Detection of mechanical injury of articular cartilage using contrast enhanced computed tomography. Osteoarthritis Cartilage 2011;19: 295e301. 31. Kulmala KAM, Pulkkinen HJ, Rieppo L, Tiitu V, Kiviranta I, Brünott A, et al. Contrast-enhanced micro-computed tomography in evaluation of spontaneous repair of equine cartilage. Cartilage 2012;3(3):235e44. 32. Kokkonen HT, Aula AS, Kröger H, Suomalainen J, Lammentausta E, Mervaala E, et al. Delayed computed tomography arthrography of human knee cartilage in vivo. Cartilage 2012;3(4):334e41. 33. Tiderius CJ, Olsson LE, Leander P, Ekberg O, Dahlberg L. Delayed gadolinium-enhanced MRI of cartilage (dgemric) in early knee osteoarthritis. Magn Reson Med 2003;49:488e92. 34. Williams A, Sharma L, McKenzie CA, Prasad PV, Burstein D. Delayed gadolinium-enhanced magnetic resonance imaging of cartilage in knee osteoarthritis: findings at different radiographic stages of disease and relationship to malalignment. Arthritis Rheum 2005;52:3528e35. 35. Owman H, Tiderius CJ, Neuman P, Nyquist F, Dahlberg LE. Association between findings on delayed gadoliniumenhanced magnetic resonance imaging of cartilage and future knee osteoarthritis. Arthritis Rheum 2008;58: 1727e30. 36. Palmer AW, Guldberg RE, Levenston ME. Analysis of cartilage matrix fixed charge density and three-dimensional morphology via contrast-enhanced microcomputed tomography. Proc Natl Acad Sci U.S.A. 2006;103:19255e60. 37. Elentuck D, Palmer WE. Direct magnetic resonance arthrography. Eur Radiol 2004;14:1956e67. 38. White LM, Schweitzer ME, Weishaupt D, Kramer J, Davis A, Marks PH. Diagnosis of recurrent meniscal tears: prospective evaluation of conventional MR imaging, indirect MR arthrography, and direct MR arthrography. Radiology 2002;222: 421e9. 39. Vande Berg BC, Lecouvet FE, Malghem J. Frequency and topography of lesions of the femoro-tibial cartilage at spiral CT arthrography of the knee: a study in patients with normal knee radiographs and without history of trauma. Skeletal Radiol 2002;31:643e9.

442 J. Hirvasniemi et al. / Osteoarthritis and Cartilage 21 (2013) 434e442 40. Turunen MJ, Töyräs J, Lammi MJ, Jurvelin JS, Korhonen RK. Hyperosmolaric contrast agents in cartilage tomography may expose cartilage to overload-induced cell death. J Biomech 2012;45:497e503. 41. Brittberg M, Winalski CS. Evaluation of cartilage injuries and repair. J Bone Joint Surg Am 2003;85-A:58e69. 42. Altman DG, Gardner MJ. Calculating confidence intervals for regression and correlation. Br Med J (Clin Res Ed) 1988;296: 1238e42. 43. Steiger J. Tests for comparing elements of a correlation matrix. Psychol Bull 1980;87:245e51. 44. Bansal PN, Stewart RC, Entezari V, Snyder BD, Grinstaff MW. Contrast agent electrostatic attraction rather than repulsion to glycosaminoglycans affords a greater contrast uptake ratio and improved quantitative CT imaging in cartilage. Osteoarthritis Cartilage 2011;19:970e6. 45. Nojiri T, Watanabe N, Namura T, Narita W, Ikoma K, Suginoshita T, et al. Utility of delayed gadolinium-enhanced MRI (dgemric) for qualitative evaluation of articular cartilage of patellofemoral joint. Knee Surg Sports Traumatol Arthrosc 2006;14:718e23. 46. Kurkijarvi JE, Nissi MJ, Kiviranta I, Jurvelin JS, 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. 47. Multanen J, Rauvala E, Lammentausta E, Ojala R, Kiviranta I, Hakkinen A, et al. Reproducibility of imaging human knee cartilage by delayed gadolinium-enhanced MRI of cartilage (dgemric) at 1.5 Tesla. Osteoarthritis Cartilage 2009;17: 559e64. 48. Burstein D, Velyvis J, Scott KT, Stock KW, Kim YJ, Jaramillo D, et al. Protocol issues for delayed Gd(DTPA)(2 )-enhanced MRI (dgemric) for clinical evaluation of articular cartilage. Magn Reson Med 2001;45:36e41. 49. Kalke RJ, Di Primio GA, Schweitzer ME. MR and CT arthrography of the knee. Semin Musculoskelet Radiol 2012;16: 57e68. 50. van Tiel J, Siebelt M, Waarsing JH, Piscaer TM, van Straten M, Booij R, et al. CT arthrography of the human knee to measure cartilage quality with low radiation dose. Osteoarthritis Cartilage 2012;20:678e85.