|Home | About | Journals | Submit | Contact Us | Français|
Assessment of subtle changes in the primary macromolecular components of cartilage, proteoglycan (PG) and collagen, is critical for the diagnosis of early stages of osteoarthritis (OA), but to date has remained a challenge. In the current study, we induced osteoarthritic cartilage changes in a rabbit model via ligament transection and medial meniscectomy and monitored disease progression with the techniques of infrared fiber optic probe (IFOP) spectroscopy, Fourier transform infrared imaging spectroscopy (FT-IRIS), and magnetic resonance imaging (MRI) microscopy. IFOP studies combined with chemometric partial least squares analysis enabled us to monitor progressive cartilage surface changes from two to twelve weeks post-surgery. FT-IRIS studies of histological sections of femoral condyle cartilage demonstrated that compared to control cartilage, the OA cartilage had significantly reduced PG content 2 and 4 weeks post-surgery, collagen fibril orientation changes 2 and 4 weeks post-surgery, and changes in collagen integrity 2 and 10 weeks post-surgery, but no significant changes in collagen content at any time point. MR microscopy studies found reduced fixed charge density (FCD), reflective of reduced PG content, in the OA cartilage compared to controls 4 weeks post-surgery. In addition, a non-significant trend towards higher apparent MT exchange rate km was found in the OA cartilage at this time point, suggesting changes in collagen structural features. These two MR findings, for FCD and km, parallel the FT-IRIS findings of reduced PG content, and altered collagen integrity, respectively. Further, MR microscopy studies at the 12 week time point cartilage found a trend towards longer T2 values and decreased anisotropy in the deep zone of the OA cartilage, consistent with increased hydration and less ordered collagen. These studies demonstrate that together, FT-IRIS and MR microscopy provide complementary data on compositional changes in articular cartilage at early stages of osteoarthritic degradation.
Osteoarthritis (OA) is an age-related disease of the joints that results in pain, frequent disability and considerable economic burden. In the United States, an average of 16 to 21 million individuals are diagnosed annually with OA, and the prevalence of the disease is further increasing because of an aging population [1–3]. The treatment of OA is hindered by the inability to identify early-stage cartilage degeneration and the lack of effective pharmaceutical interventions . Development of novel non-invasive or non-destructive detection techniques for OA, along with knowledge of specific ultrastructural changes at early stages of degeneration, would significantly enhance capabilities for early diagnosis and treatment, and facilitate the development of tissue engineering and other therapeutic interventions.
The development of OA involves progressive degeneration of articular cartilage, changes in subchondral bone and limited synovitis . Articular cartilage is comprised of a collagenous network, highly negatively charged proteoglycans (PGs), water, non-collagenous proteins and chondrocytes . Three structural cartilage zones have been defined based on morphological differences in chondrocytes and their extracellular matrix. In the superficial zone of cartilage, collagen fibrils are densely packed and oriented parallel to the articular surface . The middle, or transitional zone of cartilage contains less organized collagen fibrils and an increased concentration of PGs. In the deep zone, collagen fibrils are oriented perpendicular to the articular surface. The highest concentration of PGs and lowest water content are present in this zone. Osteoarthritic degeneration of cartilage is associated with progressive deterioration in the collagen framework, increased swelling of the tissue and loss of proteoglycan [7;8].
Animal models of OA have been developed to better understand the disease and to evaluate therapeutic interventions. Anterior cruciate ligament transection (ACLT) in the rabbit is a frequently-used and well-established technique to surgically induce joint degeneration resembling post-traumatic human OA [9–13], which occurs as a consequence of altered load distribution and joint destabilization. In these studies, osteoarthritic degeneration was evidenced by progressive changes in morphology, histopathology and physical properties of articular cartilage in the operated knees. In one study, early-stage cartilage degradation was observed as early as three weeks post-surgery .
Magnetic resonance imaging (MRI) is the most widely-used modality for non-invasive detection of pathologic changes in cartilage. Specific MRI parameters have been correlated with the tissue changes associated with OA, such as tissue swelling, PG loss and collagen network destruction [14–19]. However, only with a high field strength magnet, such as that used for in vitro studies [20–22], can detailed information on compositional cartilage changes be identified in each cartilage zone individually. The MRI-measured parameters T1 and T2 have been shown to respond to PG depletion  and changes in hydration , respectively. Other MRI microscopy studies have demonstrated orientational dependence of T2 relaxation in articular cartilage explants . The T2 anisotropy phenomenon in cartilage MRI has been shown to arise from long-range order in collagen fibril packing [24–26], and thus might be a useful tool for the detection of early changes in collagen fiber architecture associated with the development of OA . Finally, magnetization transfer (MT) imaging provides a non-invasive means of evaluating collagen content and structure in cartilage [28–30].
Fourier transform infrared (FT-IR) spectroscopy has been used extensively to study the structure and orientation of biomolecules isolated from normal and pathologic tissues [31–35]. The advantage of this technique is its ability to provide simple and often non-destructive measurements based on the unique molecular spectral signatures of tissue components. Advances in FT-IR imaging spectroscopy (FT-IRIS), a technique that couples an FT-IR spectrometer with an array detector and an optical microscope to produce an FTIR spectrum at each pixel in a defined field of view, permits mapping of spectral features and hence tissue components with a pixel resolution of 6.25 microns. In cartilage research, FT-IRIS has been applied to characterize the spatial distribution of collagen and PG, [36–40], and, with a polarizer [41;42], to determine the orientation of collagen fibrils. We have also successfully correlated changes in IR parameters with subtle cartilage degradation [39;42]. Together, these advances have made FT-IRIS a powerful tool with which to monitor cartilage degradation and to evaluate cartilage repair techniques and the composition of engineered tissues. Similarly, an infrared fiber optic probe (IFOP) has been applied in our laboratory to map the cartilage surface in vitro and in animal models [39;43], and we envision clinical application of this methodology following further development.
In the present study, two infrared spectroscopic techniques, IFOP and FT-IRIS analysis, were utilized to monitor disease progression in an ACLT-induced OA rabbit model. MRI microscopy was also performed on selected tissues to examine the correspondence of outcome measures obtained from these state-of-the-art technologies.
Surgery was performed under an IACUC-approved protocol on the right knee of 6 month old male New Zealand white rabbits to create osteoarthritic changes similar to those described in the Hulth-Tehlag Model . The surgery involved transection of the anterior and posterior cruciate ligaments and excision of the medial meniscus. A sham procedure was performed on the left hind leg to serve as a surgical control within the same animal. The rabbits were sacrificed at 0, 2, 4, 6, 8, 10 and 12 weeks post-surgery (N = 5 for each group). Animals from a group of corresponding age-matched unoperated rabbits (N = 3, except at 8 weeks, where N = 1) were sacrificed at each of these timepoints to serve as non-surgical controls.
The femoral condyles from the ACLT-operated, sham-operated and unoperated legs were excised, followed immediately by IFOP data collection. Tissues were then fixed with 80% ethanol and 1% cetylpyridinium chloride (CPC), decalcified with 10% EDTA in Tris buffer, and embedded in paraffin. Histological sections were cut at 7 μm thickness perpendicular to the articular surface and mounted onto BaF2 IR windows and glass slides for FT-IRIS and histologic analysis, respectively.
The design of the IFOP system has been described previously . Briefly, a flat-tipped, 1 mm diameter ZnS attenuated total reflectance (ATR) crystal was attached to the end of a fiber-optic bundle which was coupled to a Bruker (Billerica, MA) IR spectrometer equipped with a mercury cadmium telluride (MCT) detector. Data collection was initiated after ca. 60 seconds contact between the femoral condyle cartilage and crystal. Spectra were acquired, with 256 scans each from the weight-bearing site on each medial femoral condyle as input into the chemometric model. Spectra were collected in triplicate in order to establish a measure of spectral variability for incorporation into the model. A typical IFOP spectrum of cartilage is shown in Figure 1 with peak assignments based on previous studies of model compounds [37;40].
Chemometric methods were utilized to analyze the IFOP data. The details of the PLS method, as applied to data from human cartilage, have been previously described . Briefly, a model was developed for the current study utilizing post-surgical time as a surrogate for disease severity and based upon the PLS regression algorithm of QuantII OPUS NT software (Bruker). The infrared spectra were pre-processed for PLS analysis by performing straight line subtraction in the spectral range of 1586–999 cm−1. One hundred and thirty seven spectra were used for calibration of the PLS model, and 59 independent spectra were utilized as test input to compare the actual time post-surgery to the predictions of the model.
Transmission FT-IRIS data were acquired from histological sections of femoral condyle cartilage at 8 cm−1 spectral resolution using a Spectrum Spotlight FT-IR Imaging system (Perkin-Elmer, Bucks, UK). This system is comprised of an FTIR spectrometer coupled with a light microscope and an 8 × 2 staggered linear array detector, and allows data collection over a user-defined rectangular region at 6.25 μm pixel resolution. FT- IRIS data were also collected with a polarizer inserted into the light path to obtain information on collagen fibril orientation.
FT-IRIS images were created based on the specific vibrational absorbances for each component of interest using ISys software v3.1 (Spectral Dimensions, Olney, MD). The absorbance bands were baseline-corrected and then integrated. Previous studies have correlated collagen and PG content with the area under the protein amide I band (1598–1710 cm−1) and the integrated area of the absorbance in the range of 950–1150 cm−1, respectively [37;40]. The area under the infrared absorbance centered at 1338 cm−1 (1300–1356 cm−1), a feature attributed to side-chain CH2 vibrations, has previously been shown to decrease in intensity with collagen denaturation . This area was divided by that of the amide II band (1492–1598 cm−1) to yield a measurement of collagen integrity [39;42]. Images were calculated from data acquired with the polarizer by dividing the areas of the amide I and amide II collagen peaks for each pixel (amide I/amide II). We have previously demonstrated that this ratio provides an index of collagen fibril orientation as follows:≥2.7: fibrils parallel to the articular surface;≤1.7: fibrils perpendicular to the articular surface; 2.7–1.7: random or mixed fibril orientation . For quantitative analysis, all the pixels in the polarized FT-IRIS image were divided into three orientation categories as above, and the percentage of pixels falling into each category was calculated.
Histological slides were stained with Alcian blue and picrosirius red to demonstrate the distribution of proteoglycan and collagen, respectively. Picrosirius red, which stains collagen type I, II and III, can be used to indicate the quantity of collagen . On the Alcian blue-stained sections, dark blue corresponds to PG, red/pink to nuclei, and pale pink to cytoplasm .
MRI studies were performed on sham-operated (control; left) and operated (OA; right) femurs of three rabbits sacrificed 12 months after surgery. Experiments were performed at 4.0 ± 0.1 °C using a 400 MHz Bruker Avance DMX NMR spectrometer (Bruker Biospin, Rheinstetten, Germany) equipped with a Magnex 9.4 T/105 mm vertical bore magnet (Magnex scientific, Abingdon, UK) and a Bruker three-axis shielded Micro2.5 microimaging gradient set and 25 mm proton birdcage resonator probe.
Proton T2 was measured in a sagittal slice (thickness 0.5 mm) through the center of the medial condyle using a spin-echo sequence with a minimum TE = 9.2 ms and 16 echoes, TR=6 s, number of averages (NEX) = 8, field of view (FOV) = 0.7 × 2.4 cm, matrix size (MTX) = 256 × 256, and in-plane resolution of 27 × 94 μm. The read gradient direction was parallel to the main magnetic field, B0,. Analysis was restricted to the radial cartilage zone . Experiments were repeated with the sample rotated by approximately 55°, with the FOV increased to 1.2 × 2.4 cm, resulting in in-plane resolution of 47 × 94 μm, to enable visualization of the entire articular surface in the tilted configuration. T2 analysis was carried out in a region of interest corresponding to that analyzed in the non-tilted configuration. An anisotropy index was calculated according to:
where R2 = 1/T2. With this definition, a sample containing collagen fibrils randomly oriented relative to B0 would yield η = 0, while larger values would indicate more regular orientation.
In a separate group of six rabbits sacrificed 4 weeks post-surgery, an osteochondral plug was harvested from the weight-bearing region of the medial femoral condyle of unoperated (control; N=3) or operated (OA; N=3) legs of each animal (Figure 2). MRI microscopy was performed at 4° C with the plugs embedded in agarose gel using the hardware described above but with a 15 mm proton birdcage resonator. A 1.5 mm-thick slice was defined through the center of each plug along the anterior-posterior direction and imaged with a spin-echo sequence with FOV = 0.5 × 1.4 cm and MTX = 256 × 256, resulting in an in-plane resolution of 20 × 55 μm. The read direction was parallel to B0 and approximately perpendicular to the cartilage articular surface. Magnetization transfer (MT) experiments were performed with a presaturation pulse with offset = + 6 kHz relative to water, amplitude B1 = 12 μT, and duration tp ranging between 0.1 and 4.6 s. Values of TE = 12.8 ms, TR = 5 s and NEX = 2 were used. The MT exchange rate, km, was calculated from a fit of signal intensity as a function of saturation time tp . Matrix fixed charge density was estimated with the dGEMRIC technique , using a concentration of Gd-DTPA2− of 2 mM.
For each FT-IRIS parameter examined, a two-way ANOVA test was performed to test for the simultaneous effects of time after surgery and disease state (OA or control) on the outcome variables. Bonferroni post-hoc tests were performed for comparison of OA and control groups. Statistical significance was defined as p <0.05. All MRI data are reported as mean ± SD. For MRI data obtained from intact condyles, paired t-tests were used to compare mean T2 values and anisotropy indices between operated and contralateral legs. MT, T1 and FCD values obtained on osteochondral plugs taken from operated legs and unoperated controls animals were compared using unpaired t-tests.
The PLS correlation between the calibrated values and the actual weeks post-surgery was R2 = 94.8 using the suggested rank of 15 (Figure 3i). A prediction validation was performed against this PLS model with 59 test spectra and resulted in R2 = 93.6 with a standard error of prediction of 1.02 (Figure 3ii). This prediction gave an accuracy of ± 2 weeks for 95% percent of the test spectra.
Figure 4 shows typical FT-IRIS images and histology from medial femoral condyles obtained two weeks post-surgery from the articular surface down to the tidemark in an OA and age-matched non-surgical control animal. Qualitative differences in the spatial distribution of collagen (i: amide I, iii: picrosirius red) and proteoglycan (ii: PG/amide I, iv: Alcian Blue) between the control and OA tissues are evident; there appears to be more collagen and less PG in the OA tissue at this timepoint. Quantitative FT-IRIS data indicative of collagen content, PG content and collagen integrity averaged over the full thickness of cartilage are shown in Figure 5 at each post-surgical timepoint.
There were no significant differences in collagen content in OA versus non-surgical control at any timepoint, and no differences within each group over time (Figure 5i).
Reduced PG content, as reflected in the PG-to-collagen ratio, was found in the OA compared to the control groups at 2 and 4 weeks post-surgery by FT-IRIS, but no differences were detected within each group over time (Figure 5ii). MRI T1 and FCD values for radial zone cartilage in the osteochondral plug samples are shown in Table 1. There were no significant baseline differences in T1 between samples cut from control versus OA legs. After equilibration with Gd-DTPA2−, cartilage T1 was significantly longer in the control samples, indicating a greater matrix fixed charge density and hence greater PG content. Thus, both the MRI-derived FCD data and the FT-IRIS data show a significant reduction in PG content in the OA group at 4 weeks post-surgery.
The FT-IRIS-derived collagen integrity parameter was significantly greater in the OA groups at 2 and 10 weeks post-surgery compared to age-matched non-surgical controls (Figure 5iii), possibly indicative of a hypertrophic response consisting in part of new collagen formation or repair. There were no significant differences within each group over time. Figure 6 shows a typical T2-weighted MRI image of a plug sample from a control knee. Images from OA knees were similar in appearance. MRI magnetization transfer data for radial zone cartilage in these plugs at 4 weeks post-surgery exhibited a non-significant trend towards higher apparent MT exchange rate km in the OA group: 0.44 ± 0.17 s−1 (OA) versus 0.23 ± 0.05 s−1 (control); p = 0.16. This trend could reflect changes in collagen structural features, especially in the absence of large changes in collagen content.
The polarized FT-IRIS images indicate qualitative collagen fibril orientation changes at 2, 6 and 12 weeks post-surgery in the OA group, and are compared to polarized light microscopy (PLM) images from the same tissues (Figure 7). Collagen fibrils in the cartilage of the 2-week OA group tended to adopt a disorganized orientation, with fewer fibrils perpendicular to the articular surface as compared to the non-surgical control (Figure 7i). This phenomenon was also apparent in the PLM image. At 6 weeks post-surgery, this effect was also noted to a lesser extent (Figure 7ii), where two of five specimens showed a more random collagen fibril orientation and decreased fraction of fibrils with perpendicular orientation than controls. At 8 weeks post surgery, none of the five rabbits exhibited such effects. By 12 weeks post-surgery, severe cartilage damage, including pronounced surface fibrillation, was observed (Figure 7iii). The percentage of collagen fibrils exhibiting each of the three orientations was extracted from the polarized FT-IRIS images (Figure 8). A significantly lower fraction of pixels exhibiting perpendicular fibril orientation was observed for the 2-week and 4-week OA groups compared to controls (Figure 8iii).
MR microscopy T2 values for radial zone cartilage of the rabbit medial femoral condyles 12 weeks post-surgery are summarized in Table 2. We found a pronounced though non- significant increase in T2 with osteoarthritic degradation. This trend was more pronounced in the 0° sample orientation, for which radial zone collagen fibrils were expected to be oriented primarily parallel to the B0 field, as compared to the 55° data. Values for the anisotropy index η demonstrated a trend toward decreased anisotropy in the osteoarthritic cartilage, although again the small sample size prevented this difference from being statistically significant.
Collagen content, integrity and orientation, as well as proteoglycan content, are important determinants of cartilage function. In the current study, utilizing FT-IRIS and MRI microscopy, we investigated changes in these parameters in a rabbit model of osteoarthritis. Although similar changes have previously been found by FT-IRIS in a small number of human cartilage samples , our use of an animal model here permitted a controlled study at defined points in the progression of the disease. Further, it was possible to evaluate tissue at early stages of the disease; both FT-IRIS and MR microscopy uncovered composition changes which occurred prior to onset of gross pathologic changes.
Another recent study of experimental OA in a rabbit model found changes in type II collagen synthesis at 1, 2 and 4 weeks post-surgery . The expression of type II collagen C-propeptide, reflective of type II collagen synthesis, was found to increase over time at some sites in the OA cartilage, and to decrease at others. However, expression was always greater than in control regions. Analogous results reported in the present work are an increased FT-IRIS collagen integrity parameter in the OA tissues 2 and 10 weeks post-surgery, and a trend towards higher MT exchange rate km in the OA group at 4 weeks post-surgery as detected by MRI. All of these findings are consistent with increased type II collagen synthesis early in the development of OA, potentially indicating an active repair response. The qualitatively increased collagen content of the OA tissues compared to controls at earlier timepoints may also be attributable to this phenomenon, but did not reach statistical significance due to small sample size. Thus, it appears that the collagen integrity parameter may be a more sensitive indicator of active repair than collagen content in cartilage.
We used km rather than MTR as a probe of collagen alterations in OA, since MTR depends upon T1sat, which may itself be affected by the induction of OA. Previous studies of magnetization transfer in vitro have demonstrated that both MT ratio and km increase with increasing collagen content [28;30]. The observed trend towards increased km with OA may be due to factors other than increased collagen content. We note that km may be sensitive to a variety of collagen structural factors such as crosslinking , denaturation, fibril assembly and packing in the extracellular matrix [28;52]. In agreement with our findings Laurent et al.  utilized a 3T magnet to assess changes in vivo in a goat model of experimental OA and found that km was increased 2 weeks post surgery.
In our studies, the consistency in baseline T1 values between control and OA samples is in agreement with published data showing only a minor increase in cartilage T1 upon enzymatic 2-depletion of PG species . Measurement of T1 values after equilibration with Gd-DTPA indicated a higher proteoglycan content in the controls than in the OA samples, consistent with the substantial GAG loss from cartilage following surgical induction of experimental OA described previously , and with the FT-IRIS data in the current study.
Although MRI and FT-IRIS both provide insight into macromolecular content and structure, they are in many respects complementary. FT-IRIS is performed on dehydrated tissue sections, while MRI studies are performed on hydrated, intact tissues, including in vivo tissues in many instances. To assess PG, MRI studies indirectly measure fixed charge density through known effects of Gd-DTPA2− on tissue water relaxation, while FT-IRIS assesses a specific molecular vibration that arises directly from the PG, the sugar ring vibration [37,40]. Both approaches yield semi-quantitative information on local PG content. The differences in the manner in which PG content is determined between the two modalities could account for the discrepancy found in the percent loss of PG, which, based on FT-IRIS, was approximately 36% at the 4 week time point, while the MRI-based FCD data indicated a PG loss of 68%, nearly twice as great. Alternatively, this discrepancy could be attributable to the fact that the MRI data were obtained on an entire intact osteochondral plug, while the FT-IRIS data were obtained from a single seven-micron-thick histological section, which may not have been representative of the entire cartilage sample. Notwithstanding these differences, to our knowledge, this is the first study to demonstrate specific correspondence between parameters obtained by FT-IRIS and MRI.
In contrast to FT-IRIS, IFOP assessment is performed on hydrated tissues in a nondestructive manner. Indeed, we have previously described changes in water content in human osteoarthritic tissues . In the current study, using PLS, we were able to predict the time after surgery, a surrogate for disease severity, to within an accuracy of two weeks for the data set tested. Although this specific animal-based chemometric model cannot be directly applied to human cartilage, these results support the potential use of IFOP-based methodology for OA staging in humans, which can be performed in a minimally-invasive fashion in conjunction with arthroscopy. From a clinical perspective, the use of such a minimally-invasive technique for evaluation of cartilage quality could facilitate early intervention and assessment of repair cartilage. The ability of the chemometric analysis of IFOP results to stage early cartilage degeneration is attributable to its ability to detect subtle molecular changes in cartilage structure.
In the current study, FT-IRIS results indicated significant changes in collagen fibril orientation in the earliest stages of disease progression, but no significant changes at the later stages. Using MRI, the dependence of cartilage T2 on sample orientation was expressed as an anisotropy index η. A trend towards a smaller value for this index in the OA samples obtained 12 weeks after surgery was seen, and is somewhat in contrast to the FT-IRIS assessment of fibril orientation at this time point. However, the larger value of T2 in the OA samples is consistent with observations both in animal models [55;56] and in in vitro samples which have undergone pathomimetic enzymatic degradation [57;58]. The fact that this increase is more pronounced with fibrils oriented nominally parallel to B0 than at the magic angle suggests an alteration in fibril alignment with degradation. In addition, in each sample, radial zone T2 was greater at 55° than at 0° (data not shown). This is consistent with radial zone collagen fibril orientation being approximately perpendicular to the articular surface in both the control and OA legs. The trend towards a decrease in η with osteoarthritic degradation is consistent with a randomization of collagen fibril orientation in response to osteoarthritic degradation, a result which, to our knowledge, has not heretofore been reported in the MRI literature.
In summary, FT-IRIS and MRI microscopy together provide valuable compositional and structural information related to changes in articular cartilage at early stages of osteoarthritic degradation. These techniques together provide a powerful complementary approach that results in a comprehensive picture of molecular and biophysical changes in diseased cartilage.
This study was supported by NIH EB000744 (NPC) and the Intramural Research Program of the NIH National Institute on Aging, and utilized the facilities of the HSS Core Center for Musculoskeletal Integrity, supported by NIH AR46121. The authors wish to thank Dr. Holly Canuto (NIA) for useful suggestions.