5.2 In vivo experiments
5.2.1 ACT patients
For study IV, patients (n = 12, age 37 ±8 years) with symptomatic local cartilage defects in the femoral condyle and/or trochlea underwent ACT
A B C
Figure 5.5: Grayscale polarized microscopic (A)BF, (B)1/BF image and (C) optical density (OD) image of Safranin O–stained PGs.
surgery. The study group consisted of patients with 13 lesions in FMC (n= 6), FLC (n = 3) and trochlea (n = 4). An osteochondritis dissecans (OCD) lesion was detected in four patients. Each patient underwent a two–step ACT procedure: an initial arthroscopic evaluation and a carti-lage biopsy of a less–weight bearing healthy articular carticarti-lage area [24].
Chondrocytes were isolated from the cartilage biopsy and the number of the chondrocytes was increased in a cell culture in laboratory conditions.
After a few weeks, the chondrocytes were implanted via an injection un-derneath an autologous periosteal flap sutured over the debrided cartilage lesion [24, 119] (figure 5.6).
5.2.2 MRI measurements
The patients were examined at 10 to 15 months after the surgery by MRI at 1.5 T (GE Signa 1.5 T, Milwaukee, WI) using a quadrature transmit and receive imaging extremity coil. First a clinical MRI exam of the knee joint was conducted, including aPD–weighted fast spin echo (FSE) series with fat suppression in coronal (TR= 3140 ms, effectiveTE= 25 ms, echo train length (ETL) of 9, 256x224 imaging matrix, 0.63x0.71 mm in–plane resolu-tion), sagittal (TR= 3500 ms, effectiveTE= 41 ms,ETL= 8, 256x256 mat-rix, 0.63x0.63 mm in–plane resolution) and axial directions (TR = 2800 ms, effective TE= 25 ms,ETL= 8, 256x224 matrix, 0.63x0.71 mm in–plane re-solution), coronalPD–weighted FSE series without fat suppression (TR=
5.2In vivoexperiments 49
A B C
Figure 5.6: Autologous chondrocyte transplantation. (A) Degenerated ar-ticular cartilage was removed (B), autologous periosteal flap was sutured over the debrided cartilage lesion and (C) chondrocytes were injected under it.
2500 ms, effectiveTE= 21 ms,ETL= 6, 256x256 matrix, 0.63x0.63 mm in–
plane resolution). The field–of–view inPD–weighted images was 16–cm.
A sagittalT1–weighted spin echo series was conducted withTR= 600 ms, effective TE= 9 ms, 512x320 matrix, 32–cm FOV, 0.63x1.00 mm in–plane resolution. In every imaging series, the slice thickness was 3–mm.
After the clinical knee examination T2 relaxation time mapping was performed. A series of sagittal and coronal single–slice FSE measure-ments from the center of the graft were conducted with varying echo times (TR= 2000 ms, six effectiveTEs between 18 and 110 ms,ETL= 9, 512x512 matrix, 16–cm FOV, 0.31x0.31 mm in–plane resolution, 3–mm slice).
Following theT2, an intravenous injection of 0.2 mM/kg Gd-DTPA2−
(Schering AG, Berlin, Germany) was given to permit the dGEMRIC expe-riments. After a 2–h delay, the T1 relaxation time was determined at the location of theT2 measurement, using a series of single slice inversion re-covery FSE measurements in the sagittal and coronal planes (TR = 1800 ms,ETL= 9,TE= 17 ms, sevenTIs between 50 and 1650 ms, 512x512 ma-trix, 16–cm FOV, 0.31x0.31 mm in–plane resolution, 3-mm slice). Coronal T2 and dGEMRIC data were available for nine grafts. The sequences in use are presented in figure 5.3.
5.3 Data analysis
In study I, the non-parametric Friedman test and the Friedman post hoc test for several dependent samples were utilized in order to assess the to-pographical variations in MRI and biomechanical parameters (SPSS Soft-ware; SPSS Inc., Chicago, IL). The Pearson’s linear correlation test was used to analyze the interrelationships between the MRI and biomechani-cal parameters. Bulk values of each parameter were biomechani-calculated as the mean value of the profiles, whereas the surface values for MRI experiments con-sisted of the averages calculated from the 3x3 pixels on the cartilage sur-face.
Study II concentrated on the comparison of PLM– andT2–profiles (fi-gure 6.4). The profiles were equalized by downsampling the BF data to match the MRI resolution. A truncation from deep cartilage was per-formed to equalize the profile vector lengths [134]. A pixel–by–pixel ana-lyzed linear Pearson correlation was performed between the resampled 1/BF profiles and surface-matched depthwise (from cartilage surface to subchondral bone) T2 and T2Gd datasets for each sample, and for pooled data (SPSS software, SPSS inc., Chicago, IL). The Wilcoxon signed ranks test was used to compare the difference betweenT2andT2Gd.
From 1/BF and T2, the profiles thickness of histological zones were determined and compared. The zone–boundaries were determined for each rising/declining edge as the half–maximum value, a method adapted from Xiaet al., 2001 [198]. A custom-made MATLAB program (MathWorks Inc. , Natick, MA, USA) was used for the calculations. The Bland–Altman plot was used to assess the agreement of MRI and PLM techniques for the determination of the locations of the lamina boundaries [18]. The Kruskal–
Wallis post hoc test was applied in order to investigate the significance of differences in zone thicknesses. To analyse the reproducibility of the measurements, the coefficient of variation (CV%) for depthwise and deep tissue data was calculated [58].
In study III, the depthwise MR and OD data were compared. In order to enable a sample–by–sample comparison, the profiles were linearly in-terpolated to 10µm spatial resolution and truncated from the deep tissue to the shortest length. Bulk values of each parameter were calculated as the mean value of these profiles. Using the bulk values of the parame-ters and bulk water content, statistical comparisons between the different sites in the knee joint were conducted, applying Friedman post hoc test.
Pearson’s linear correlation and the first order partial correlation test were used to demonstrate the associations between the parameters.
In study IV, T2 relaxation time maps were calculated by means of a
5.3 Data analysis 51 mono–exponential two–parameter fitting in MATLAB (version 7.0.4, Math-works Inc., Natick, MA) using all time points. T1 relaxation time maps were calculated using a three–parameter fitting. Sagittal and coronal va-lues were determined from the manually segmented full thickness ROIs, matched in location and size (area 6–9 mm2, mean 7±1 mm2) from the cen-tral portion of the graft. Superficial and deep half of the tissue in each graft were also studied. Relaxation time values were similarily determined for control tissue, i.e. in the sagittal plane, the cartilage adjacent to the graft, with normal signal intensity, was chosen. In the proton density–weighted images, the grafts in the posterior part of the joint had anterior control ROIs, and the grafts positioned at the anterior part had a control ROI at the posterior part. In the coronal orientation, control ROIs were chosen from the respective site of the contralateral side of the joint.P D–weighted images were categorized based on the signal intensity as hypo–, iso– or hyperintense, as compared to the surrounding cartilage. Relaxation time values were compared with the non–parametric Wilcoxon’s signed ranks test (SPSS Inc., Chicago, IL, USA). The Pearson correlation coefficient was determined for T2 and dGEMRIC values for control and graft tissue. A Kruskal–Wallis test was used to determine the dependence between the P D–weighted signal intensity and the quantitative MRI parameters of the grafts. Relaxation time values of grafts at different anatomical locations (i.e.femur and trochlea) were compared using the Mann–Whitney test.