Sodium and T1rho MRI for molecular and diagnostic imaging of articular cartilage

Abstract

In this article, both sodium magnetic resonance (MR) and T1rho relaxation mapping aimed at measuring molecular changes in cartilage for the diagnostic imaging of osteoarthritis are reviewed. First, an introduction to structure of cartilage, its degeneration in osteoarthritis (OA) and an outline of diagnostic imaging methods in quantifying molecular changes and early diagnostic aspects of cartilage degeneration are described. The sodium MRI section begins with a brief overview of the theory of sodium NMR of biological tissues and is followed by a section on multiple quantum filters that can be used to quantify both bi-exponential relaxation and residual quadrupolar interaction. Specifically, (i) the rationale behind the use of sodium MRI in quantifying proteoglycan (PG) changes, (ii) validation studies using biochemical assays, (iii) studies on human OA specimens, (iv) results on animal models and (v) clinical imaging protocols are reviewed. Results demonstrating the feasibility of quantifying PG in OA patients and comparison with that in healthy subjects are also presented. The section concludes with the discussion of advantages and potential issues with sodium MRI and the impact of new technological advancements (e.g. ultra-high field scanners and parallel imaging methods). In the theory section on T1rho, a brief description of (i) principles of measuring T1rho relaxation, (ii) pulse sequences for computing T1rho relaxation maps, (iii) issues regarding radio frequency power deposition, (iv) mechanisms that contribute to T1rho in biological tissues and (v) effects of exchange and dipolar interaction on T1rho dispersion are discussed. Correlation of T1rho relaxation rate with macromolecular content and biomechanical properties in cartilage specimens subjected to trypsin and cytokine-induced glycosaminoglycan depletion and validation against biochemical assay and histopathology are presented. Experimental T1rho data from osteoarthritic specimens, animal models, healthy human subjects and as well from osteoarthritic patients are provided. The current status of T1rho relaxation mapping of cartilage and future directions is also discussed.

Figure 1

The extracellular matrix (ECM) of cartilage.

Figure 2

The aggrecan proteoglycan macromolecule (A) comprising hyaluronic acid (HA), keratan sulfate (KS) and chondroitin sulfate (CS). The molecular structures of chondroitin 6-sulfate and keratan sulfate moieties are shown in (B).

Figure 3

Arrangement of collagen fibers across different layers of the cartilage.

Figure 4

Energy level diagram of spin 3/2 nucleus. The frequency difference between adjacent energy levels is the same, ω₀ (A). It represents the isotropic solution state, where quadrupolar interaction is averaged to zero. (B) Nonzero static quadrupolar interaction shifts the energy levels, producing three spectral peaks. This represents the situation in solids, liquid crystals and oriented macromolecules. The frequency separation between adjacent energy level are given by: (ω₀ − ω_Q), ω₀ and (ω₀ + ω_Q).

Figure 5

Pulse sequence for multiple quantum filtering. τ, δ and t are preparation, evolution and detection times, respectively. Tensor components during each time period represent the possible coherences. Appropriate phase cycling of the RF pulses is used to select either double or triple quantum coherences.

Figure 6

Simulated SQ, TQF and DQF-MA spectra [from eqns (21)–(23)] for different quadrupolar splitting frequencies (ν_Q = ω_Q/2π). The evolution time τ = 2 ms, T_2fast = 1 ms, T_2slow = 15 ms. At low ω_Q values greater than 0, clear splittings are absent and experimental DQF-MA spectra can be fitted to extract ω_Q directly.

Figure 7

Sodium multiple quantum coherences from articular cartilage: (A) single quantum, (B) double quantum coherence $(T_2^2+T_2^3)$ and (C) triple quantum coherence $(T_3^3)$. Pronounced negative lobes in double quantum coherence $(T_2^2+T_2^3)$ compared with triple quantum coherence is due to the nonaveraged quadrupofar interaction. (D) Normalized triple quantum spectral intensity from human articular cartilage in vivo plotted as a function of preparation time. The solid line is the fitted to the TQF signal expression (142).

Figure 8

(A) Axial slice of sodium image of a healthy human knee. The high sodium content in articular cartilage is clearly visible. The signal profile across the cartilage is shown in (B) (27).

Figure 9

Quantitation and calibration of sodium concentration in articular cartilage. Sodium image of bovine cartilage (A). Trypsin-treated region has low [Na] due to reduced GAG. Circular sodium agarose phantoms that were used to quantify [Na] are also seen in this figure. A plot of FCD measured from sodium MR and that obtained from DMMB assay (B) shows a strong linear correlation between these measurements (29).

Figure 10

Sodium concentration maps of human patellar cartilage specimens obtained following knee replacement surgery. The top image is from a healthy cartilage while the bottom image is that from an osteoarthritic patient. The scale bar indicates sodium concentration in mm.

Figure 11

Sodium images of the human knee joint in vivo acquired in the axial plane. Cartilage in the patellar–femoral joint is visible in yellow in the image on the left. The scale bar indicates sodium concentration in mm. The image on the left was obtained on a healthy volunteer where uniform high sodium content was observed. The right image is from a symptomatic osteoarthritic subject, demonstrating heterogeneous and low [Na] that reflects a loss of GAG in this subject (30).

Figure 12

Sodium concentration maps of saline and IL-1β-treated knee joints in swine. The maps shown were acquired from a single specimen, but are representative of the data acquired from all six animals. The resolution of the in vivo and ex vivo maps were kept identical to allow direct comparison (154). [Na] and FCD from cartilage of saline-treated and IL-1β-treated joints of all six animals (B) demonstrates a significant reduction in FCD from treated cartilage.

Figure 13

Pulse sequence for spin-locking magnetization in the transverse plane. Initially, a π/2 pulse flips the longitudinal magnetization into the transverse plane. The open rectangle pulse represents the spin lock (SL) pulse and TSL and B₁ are its duration and amplitude, respectively. The dotted line represents the decay of the magnetization in the absence of spin-locking and is governed by time constant, $T_2^{*}$. However, if the magnetization is spin-locked, it decays according to T_1ρ (solid line) for the duration of time TSL. But after the SL pulse the decay is again $T_2^{*}$, during which time the signal may be acquired.

Figure 14

Radiofrequency pulse cluster for T_1ρ pre-encoding (left). TSL and B₁ are the length and amplitude of the spin-lock (SL) pulse, respectively. Vector diagram (below) of the evolution of the magnetization during the spin-lock pulse cluster.

Figure 15

The pulse sequence for acquiring T_1ρ-weighted true FISP (or TRUFI) images. The steady-state magnetization is stored in the z-direction by the last α/2 pulse of the TRUFI sequence and then recalled by the T_1ρ preparatory cluster and returned to the z-axis by the second hard π/2 pulse. A string crusher gradient (indicated by the dotted box) in then applied at that point to destroy any residual magnetization in the transverse plane.

Figure 16

Sagittal T_1ρ-TRUFI (A) and T_1ρ-TSE (B) maps (in color) are overlaid on a grayscale image of the femor–tibial cartilage of a 22-year-old healthy volunteer. Also shown is an axial T_1ρ-TRUFI map from another healthy volunteer’s patellar–femoral cartilage (C). The color bar on the right indicates T_1ρ values in milliseconds. Although reduced values are observed in parts of the T_1ρ-TSE map, the average T_1ρ is not significantly different between the T_1ρ-TRUFI (39.4 ms) and T_1ρ-TSE (38.4 ms) maps. A region of elevated T_1ρ is evident on the lateral facet of the patellar cartilage in (C) and could be an indication of early-stage cartilage degeneration.

Figure 17

Pulse-acquire proton NMR spectrum of CH₂Cl₂ in a nematic solvent (doublet of 4.2 kHz, arising from the dipolar interaction between the two protons). The distorted baseline is due to the broad NMR spectrum of the liquid crystal (221).

Figure 18

Evolution of dipolar doublet as function of the spin-lock period duration; spin-lock amplitude: w₁= 4 kHz. These experimental results were obtained by subtracting the anti-phase doublet signal. Note the initial rapidly damped oscillations and the mono-exponential decay for the major part of the evolution curve (221).

Figure 19

T_1ρ-weighted signal intensity from a bovine cartilage as a function SL length is shown. The ω₁ used for this study is 250 Hz. The inset shows the expanded version of the initial portion of the signal. The oscillation pattern is characteristic of the presence of RDI in the tissue (240).

Figure 20

T₂- and T_1ρ-weighted images of bovine articular cartilage. In T₂-weighted images, laminae of alternating high and low signal intensity pattern are evident. In T_1ρ-weighted images, the laminar appearance is gradually reduced with increasing ω₁, making the signal more homogeneous across the cartilage (222).

Figure 21

T_1ρ dispersion in bovine cartilage; the cartilage surface was oriented parallel to B₀ and at the magic angle ~55°. The data at spin-lock frequency of zero correspond to the respective T₂ values (222).

Figure 22

(A) A T₂-weighted (i) and corresponding T_1ρ-weighted image (ii) and T_1ρ map (iii) of a slice through bovine articular cartilage. The T_1ρ dispersion collected for spin-lock amplitude (B₁) from 0 to 8 kHz is shown in (B). where a twofold increase in T_1ρ was observed with increasing B₁ amplitude, demonstrating T_1ρ dispersion in cartilage (230).

Figure 23

These figures show a plot of 1/T₂ vs PG and 1/T_1ρ vs PG loss from a group of bovine cartilage patellae subjected to serial depletion of PG. The solid line indicates the linear fit to the experimental data. Although T₂ did change with PG content, there was no clear trend (r² = 0.01, p < 0.7). T_1ρ data from the same set of cartilage specimens, however, demonstrated a strong correlation (r² = 0.9, p < 0.001) between changes in PG and 1/T_1ρ(226).

Figure 24

Comparison of T₂, and T_1ρ maps of control and 40% of PG depleted bovine patellae. (A) Control T₂ map; (B) control T_1ρ map; (C) 40% PG depleted T₂ map; (D) 40% PG depleted T_1ρ map. The color scale bar shows the relaxation numbers from 0 to 256 ms (25).

Figure 25

T_1ρ maps of representative bovine cartilage specimens in culture media. The treated group represents the treatment with 30 ng/mL of IL-1β over a 10-day period. Significant elevation in T_1ρ maps of specimens compared with control specimens is evident (241).

Figure 26

Linear regression plots of average data from each group of (a) H_A vs PG, (b) log₁₀k₀ vs PG, (c) bulk R_1ρ vs PG, (d) mid-zone R_1ρ vs PG, (e) H_A vs R_1ρ and (f) log₁₀k₀ vs R_1ρ. Open data points represent control groups and closed data points represent IL-1β-treated groups. The correlation coefficient (R²) is listed on each corresponding plot. Error bars on both axes represent the SD of each measurement (241).

Figure 27

Correlation of normalized FCD and R_1ρ data obtained from the trypsin treated model (A). The slope of the linear regression was 0.48 ± 0.004 with r² > 0.75 (p < 0.001). The data are arranged in four clusters representing the four groups of trypsin concentration. The cluster about zero, including negative values, represents the distribution of the mean of the control data. Correlation of normalized FCD and R_1ρ data obtained from the natural OA model (B). The slope of the linear regression was 0.51 ± 0.06 with r² > 0.85 (p < 0.001) (177).

Figure 28

Comparison of sodium concentration and T_1ρ maps of a representative ex vivo patella from an osteoarthritic joint. The sodium map reveals a distinct reduction in FCD on the lateral side of the patella (ROI A) indicating a reduction of PG content. Elevation of T_1ρ in ROI A is observed as an analogous measure of PG loss (177).

Figure 29

T_1ρ maps of saline- and IL-1β-injected patellae of a Yorkshire pig (A) in vivo (top row) and the same ex vivo (bottom row). Color-coded T_1ρ maps corresponding to cartilage are overlaid on the original T_1ρ-weighted MR image, indicating elevated T_1ρ in each IL-1 treated case (54). Also shown is a plot of 1/T_1ρ measured in vivo and FCD measured by post-mortem PG assay from animals (B).

Figure 30

Representative T_1ρ-weighted images of a guinea pig (age ~4 months) and corresponding color-coded T_1ρ maps (overlaid on the T_1ρ-weighted images). (A) An image of the knee joint showing the T_1ρ map of patellar cartilage in color. The average T_1ρ of patellar cartilage was measured at 57 ± 5 ms. (B) A sagittal slice of the same animal’s knee joint showing both femoral and tibial cartilage.

Figure 31

In vivo T₂- and T_1ρ-weighted images from a healthy human knee joint. The imaging parameters were TR 3s, FOV = 10cm × 10cm, slice thickness = 3 mm, matrix = 256 128 (a) T₂-weighted image, TE = 16 ms (b) T_1ρ-weighted image, TE +TSL = 16 ms, B₁ = 500 Hz.

Figure 32

T_1ρ-weighted image (gray-scale) and corresponding overlaid color map of a clinically diagnosed OA subject’s knee joint. This patient did not display OA in radiographic images. The region indicated by the oval in the patellar–femoral cartilage indicates a region of elevated T_1ρ (>60 ms) and could be a sign of early OA.

Figure 33

T_1ρ-weighted images of a symptomatic OA subject without any radiographic OA. (A) and (B) represent the images obtained with an effective weighting of 45 and 60 ms respectively. Elevated signal intensity demonstrates a lesion on cartilage.

Figure 34

Preliminary results from an OA subject arthroscopically diagnosed with grade I chondromalacia in the lateral facet of the patella. The left-hand side figure shows the three-dimensional T_1ρ relaxation map of patellar cartilage. The color scale shows a volume-rendered representation of the T_1ρ numbers. The image on the right shows a slice of the T_1ρ map at the position indicated by a line on the T_1ρ surface (on the left) overlaid on the patellar cartilage of the proton density-weighted image. The dashed elliptical region on image on the left is the arthroscopically confirmed region of chondromalacia.

Figure 35

Colour-coded T_1ρ maps overlaid on SPGR images. Left: patellar cartilage; middle: anterior femoral (trochlea) cartilage; right: posterior femoral cartilage, (a) a healthy volunteer, male, 30; (b) a patient with early OA, female, 27. The T_1ρ values were 40.05 ± 11.43 ms in the volunteer and 50.56 ms ± 19.26 ms in the patient, respectively.

Figure 36

Correlation between T_1ρ and T₂ values in 12 subjects (five healthy controls and seven patients with OA). Although the correlation is significant, there is a nonpoint-to-point relationship, that is, patients with similar T₂ values may show different T_1ρ values, as shown in the vertical dashed box, or vice versa, as shown in the horizontal dashed box. The dashed line shows a clear-cut difference of the average T_1ρ between healthy controls and OA patients, while there is a significant overlap of average T₂ values. A significant difference was found in average T_1ρ (p = 0.003) but not in T₂ values (p = 0.2002) between healthy controls and patients (183).