Convincing evidence for magic angle less‐sensitive quantitative T_1ρ imaging of articular cartilage using the 3D ultrashort echo time cones adiabatic T_1ρ (3D UTE cones‐AdiabT_1ρ) sequence

Abstract

Purpose

To investigate the magic angle effect in three‐dimensional ultrashort echo time Cones Adiabatic T_1ρ (3D UTE Cones‐AdiabT_1ρ) imaging of articular cartilage at 3T.

Methods

The magic angle effect was investigated by repeated 3D UTE Cones‐AdiabT_1ρ imaging of eight human patellar samples at five angular orientations ranging from 0° to 90° relative to the B₀ field. Cones continuous wave T_1ρ (Cones‐CW‐T_1ρ) and Cones‐ ${\mathrm{T}}_2^{\ast }$ sequences were also applied for comparison. Cones‐AdiabT_1ρ, Cones‐CW‐T_1ρ and Cones‐ ${\mathrm{T}}_2^{\ast }$ values were measured for four regions of interest (ROIs) (10% superficial layer, 60% transitional layer, 30% radial layer, and a global ROI) for each sample at each orientation to evaluate their angular dependence.

Results

3D UTE Cones‐AdiabT_1ρ values increased from the radial layer to the superficial layer for all angular orientations. The superficial layer showed the least angular dependence (around 4.4%), while the radial layer showed the strongest angular dependence (around 34.4%). Cones‐AdiabT_1ρ values showed much reduced magic angle effect compared to Cones‐CW‐T_1ρ and Cones‐ ${\mathrm{T}}_2^{\ast }$ values for all four ROIs. On average over eight patellae, Cones‐AdiabT_1ρ values increased by 27.2% (4.4% for superficial, 23.8% for transitional, and 34.4% for radial layers), Cones‐CW‐T_1ρ values increased by 76.9% (11.3% for superficial, 59.1% for transitional, and 117.8% for radial layers), and Cones‐ ${\mathrm{T}}_2^{\ast }$ values increased by 237.5% (87.9% for superficial, 262.9% for transitional, and 327.3% for radial layers) near the magic angle.

Conclusions

The 3D UTE Cones‐AdiabT_1ρ sequence is less sensitive to the magic angle effect in the evaluation of articular cartilage compared to Cones‐ ${\mathrm{T}}_2^{\ast }$ and Cones‐CW‐T_1ρ.

1. INTRODUCTION

Knee osteoarthritis (OA) affects millions of people worldwide. One of the most significant changes that can be observed early in the development of the disease is the depletion of proteoglycans (PGs) in articular cartilage. Conventional MRI sequences are widely used for morphological characterization of cartilage loss in the later stages of OA, but have exhibited little success in the characterization and diagnosis of early OA. Spin lattice relaxation in the rotating frame, namely T_1ρ, has been proposed as an alternative to conventional MRI sequences for detecting biochemical changes in cartilage ¹ , ² , ³ ; more specifically, a number of studies have proposed T_1ρ as a biomarker for PG loss in both bovine cartilage samples and in patients with OA. ⁴ , ⁵ , ⁶ However, a major confounding factor in utilizing continuous wave T_1ρ (CW‐T_1ρ) imaging for articular cartilage is the magic angle effect. ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ The ordered collagen fibers in cartilage are subject to dipole‐dipole interactions that are modulated by the term 3cos2(θ)‐1, where θ is the angle between the fiber orientation and the main magnetic field B₀. Recent studies suggest that CW‐T_1ρ is subject to a strong magic angle effect, demonstrating a 100% or more increase in T_1ρ values when θ is changed from 0° to 55°. ⁸ , ⁹ , ¹⁰ This magic angle‐induced increase can far exceed changes caused by degeneration, which are typically in the range of 10%‐30%, ¹² thereby complicating clinical diagnosis and treatment monitoring.

To address this challenge, researchers have proposed trains of adiabatic full passage (AFP) pulses to generate adiabatic T_1ρ (AdiabT_1ρ) relaxation. ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ Previous studies have shown that AdiabT_1ρ is less sensitive to the magic angle effect compared to CW‐T_1ρ relaxation in bovine cartilage, although both show strong orientation anisotropy dependence on the spin‐lock power. ¹⁷ , ¹⁸ The regular AdiabT_1ρ sequence is based on conventional Cartesian data acquisition with an echo time (TE) of several milliseconds or longer, which, while short enough for high resolution imaging and AdiabT_1ρ quantification of certain layers of articular cartilage, is too long to image many knee joint tissues and tissue components with short ${\mathrm{T}}_2^{\ast }$ s such as the deepest radial layers of articular cartilage, menisci, ligaments, tendons, and bone. ²⁰

Meanwhile, OA is a whole‐organ disease, which involves damage to not only the superficial and transitional layers of cartilage, but also to the deep radial layer of cartilage, menisci, ligaments, tendons, subchondral bone‐cartilage interface, and subchondral bone. ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ The failure of any one of the joint tissues has the potential to affect others, contributing to overall failure of the joint as a whole. ²⁸ This confluence of disease mechanics and imaging capability demonstrates the crucial need for magic angle‐insensitive quantitative imaging, such as AdiabT_1ρ imaging, of all the major knee joint tissues to achieve more robust and systematic assessment of OA.

Ultrashort TE (UTE) sequences with TEs less than 0.1 ms have been proposed for high resolution morphological imaging of both short and long T2 tissues in the knee joint. ²⁹ , ³⁰ , ³¹ 3D UTE with Cones sampling has been shown to be highly time‐efficient for volumetric imaging of short T2 tissues. ³² Quantitative UTE imaging techniques, including CW‐T_1ρ mapping and ${\mathrm{T}}_2^{\ast }$ mapping, have also been developed. ³¹ , ³³ , ³⁴ However, UTE CW‐T_1ρ and ${\mathrm{T}}_2^{\ast }$ mapping techniques are highly sensitive to the magic angle effect. ⁹ , ³⁵ Since AdiabT_1ρ is less sensitive to the magic angle effect ¹⁷ , ¹⁸ and UTE Cones imaging is highly time‐efficient, ³² the combination of 3D UTE Cones data acquisition with AdiabT_1ρ preparation (3D UTE Cones‐AdiabT_1ρ) could provide clinically feasible, magic angle‐less sensitive imaging of various knee joint tissues. ³⁶

In this study, we aimed to evaluate the magic angle susceptibility of the 3D UTE Cones‐AdiabT_1ρ sequence in imaging cadaveric human patellar cartilage samples on a clinical 3T scanner. Because the different layers in articular cartilage (i.e., superficial, transitional, deep radial, and calcified layers) have distinct collagen fiber structures, it is important to understand whether 3D UTE Cones‐AdiabT_1ρ values of articular cartilage as a whole, as well as values of the cartilage within different layers, are less sensitive to angular orientation before applying the 3D UTE Cones‐AdiabT_1ρ sequence in clinical evaluation. Our study systematically investigated the magic angle effect by repeated high‐resolution 3D UTE Cones‐AdiabT_1ρ imaging of eight human patellar cartilage samples at five angular orientations ranging from 0° to 90° relative to the B₀ field. Conventional 3D UTE Cones‐CW‐T_1ρ and Cones‐ ${\mathrm{T}}_2^{\ast }$ sequences were also applied for comparison. If confirmed to be magic angle‐less sensitive, the 3D UTE Cones‐AdiabT_1ρ sequence may greatly improve robustness in quantitative systematic evaluation of knee joint degeneration.

2. METHODS

2.1. Human patellar cartilage samples procurement

Our institutional Human Research Protections Program approved the procurement and use of human tissues in our study. Eight cadaveric human patellar samples were harvested from freshly frozen knee specimens (eight donors aged 38‐61 years, mean age 49.1 ± 9.7 years; three males, five females) provided by a non‐profit whole‐body donation company (United Tissue Network, Phoenix, AZ). After harvesting, a transverse slab of normal‐appearing, approximately 8‐mm thick tissue was cut from each specimen using a low‐speed diamond saw (IsoMet 1000, Buehler) and stored in phosphate‐buffered saline (PBS) for 4 hours at room temperature for rehydration. During MRI scanning, the patellar cartilage samples were placed in a 30‐ml syringe filled with Fomblin (Fomblin, Ausimont, Thorofare, NJ) to minimize dehydration and susceptibility artifacts at air‐tissue interfaces.

2.2. MR data acquisition

The 3D UTE Cones‐AdiabT_1ρ sequence was implemented on a clinical 3T MR750 scanner (GE Healthcare Technologies, Milwaukee, WI). The sequence employed an even number of adiabatic inversion recovery (NIR) pulses (Silver‐Hoult pulse shape, duration = 6.048 ms, peak power = 17 µT) followed by regular 3D UTE Cones imaging, which included a short rectangular pulse excitation followed by Cones sampling. ³² Following each adiabatic T_1ρ preparation, fast Cones data acquisition was performed using a number of spokes (Nsp) with an equal time interval τ. This multi‐spoke approach is expected to speed data acquisition by a factor of Nsp. The spin lock time (TSL) is defined as the total duration of the train of adiabatic IR pulses, that is, TSL = NIR × Tp (Tp is duration of a single adiabatic IR pulse). AdiabT_1ρ values were quantified through exponential fitting of 3D UTE Cones‐AdiabT_1ρ signals as a function of TSLs, as reported in our prior study. ³⁶ Accurate T1 measurement is needed for T_1ρ calculation because of the use of a relatively short repetition time (TR). 3D UTE Cones actual flip angle imaging (AFI) was used for mapping of B1 inhomogeneity, which, together with a variable flip angle (VFA) method (3D UTE Cones AFI‐VFA), provided for accurate T1 mapping of articular cartilage, including the superficial layers with relatively long T2s and the deep radial layers with relatively short T2s. ³⁷ , ³⁸ Features of the various 3D UTE Cones sequences used in this study are shown in Figure 1.

FIGURE 1.

Quantitative 3D UTE Cones sequences include the basic 3D UTE Cones sequence (A), which uses a short rectangular pulse for signal excitation followed by 3D spiral sampling with a minimal nominal TE of 32 µs and conical view ordering (B), the 3D UTE Cones‐actual flip angle imaging (AFI) sequence with dual‐TR acquisitions for B1 mapping (C), the conventional 3D UTE Cones sequence with a single TR for T1 measurement with the variable flip angle (VFA) or variable repetition time (VTR) method (D), and the 3D UTE Cones‐AdiabT_1ρ sequences for AdiabT_1ρ measurement (E). To speed up data acquisition, multiple spokes (Nsp) were sampled after adiabatic T_1ρ preparation

Typical imaging parameters included a field of view (FOV) of 5 × 5 cm², a slice thickness of 0.8 mm, and a receiver bandwidth (BW) of 100 kHz. Other sequence parameters were: 1) Cones‐AFI ³⁷ : TR1/TR2 = 20/100 ms, flip angle (FA) = 45°, acquisition matrix = 192 × 192 × 40, scan time = 9 min 50 s; 2) Cones‐VFA ³⁷ : TR = 20 ms, FA = 4°, 7°, 10°, 15°, 20°, 25° and 30°; acquisition matrix = 192 × 192 × 40, scan time = 16 min 20 s; 3) Cones‐AdiabT_1ρ ³⁶ : TR = 1000 ms, FA = 8°, acquisition matrix = 192 × 192 × 40, Nsp = 15, NIR = 0, 4, 8, 12 and 16, each with a scan time of 7 min 30 s; 4) Cones‐CW‐T_1ρ ³⁴ : TR = 1000 ms, FA = 8°, acquisition matrix = 192 × 192 × 40, Nsp = 15, spin‐lock power = 500 Hz, TSL = 0, 5, 10, 15 and 20 ms, each with a scan time of 7 min 30 s; 5) Cones‐ ${\mathrm{T}}_2^{\ast }$: TR = 45 ms, FA = 15°, acquisition matrix = 192 × 192 × 40, fat saturation, one set of multi‐echo acquisitions (TEs = 0, 4.4, 8.8, 13.2 and 22 ms) with a scan time of 11 min 8 s. The imaging protocol was repeated five times, each with a different orientation (0°, 30°, 55°, 75° and 90° relative to B₀). The rotating scheme is shown in Figure 2 (based on localizer). A transmit/receive 1‐inch diameter birdcage coil was used for signal excitation and reception.

FIGURE 2.

The rotating scheme in the magic angle study of cadaveric human patellar cartilage samples. Patellar cartilage samples were rotated 0°, 30°, 55°, 75° and 90° relative to the B₀ field

2.3. Image analysis

3D UTE Cones‐AdiabT_1ρ, Cones‐CW‐T_1ρ and Cones‐ ${\mathrm{T}}_2^{\ast }$ datasets acquired at five different angular orientations were first manually aligned using ImageJ software, then automatically registered with FLIRT (Functional MRI of the Brain’s Linear Image Registration Tool) software using six parameter rigid body model and correlation ratio as the cost function. ³⁹ T1 is known to be magic angle‐independent and was acquired only once. T1 correction was done per the protocols of prior publications. ³⁴ , ³⁶ The DICOM images were analyzed with MATLAB 2017b (The MathWorks, Natick, MA). Two radiologists individually drew four ROIs on the images of each patellar cartilage sample (i.e., 10% superficial, 60% transitional, 30% deep radial, and a global ROI comprising the entire region; the depth for each region was defined per prior publications ⁴⁰ ), and regional analysis was applied to each sample. We used a semi‐automated, MATLAB program developed in‐house to copy and paste ROIs to the registered images, a technical approach which ensured that ROIs were located identically on images that had been obtained at different angles and sequences. Single‐component model was applied to fit T1, T_1ρ, AdiabT_1ρ and ${\mathrm{T}}_2^{\ast }$ according to prior published procedures. ³¹ , ³² , ³³ , ³⁵ , ³⁶ , ³⁷ The angular dependence of each biomarker was analyzed. Intraclass correlation efficient (ICC) was used to evaluate consistency between the two radiologists.

3. RESULTS

Figure 3 shows representative images from 3D UTE Cones‐AdiabT_1ρ imaging, Cones‐CW‐T_1ρ imaging and Cones‐ ${\mathrm{T}}_2^{\ast }$ imaging of the same patellar sample oriented 0° and 55° relative to the B₀ field. The fastest signal decay was observed in 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$ imaging, especially for the transitional and deep radial layers of patellar cartilage, while the slowest signal decay was observed in 3D UTE Cones‐AdiabT_1ρ imaging. Furthermore, 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$ imaging showed the greatest signal difference between 0° and 55° relative to the B₀ field, whereas 3D UTE Cones‐AdiabT_1ρ imaging showed the smallest signal difference between those two orientations.

FIGURE 3.

3D UTE Cones‐AdiabT_1ρ imaging of a patellar sample oriented parallel (1st row) and 55° (2nd row) relative to the B₀ field, with TSLs of 0 ms (A, a), 24 ms (B, b), 48 ms (C, c), and 72 ms (D, d). Regular 3D UTE Cones‐T_1ρ imaging of the same patellar sample oriented parallel (3rd row) and 55° (4th row) relative to the B₀ field, with TSLs of 0 ms (E, e), 5 ms (F, f), 10 ms (G, g), and 15 ms (H, h). 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$ imaging of the same patellar sample oriented parallel (5th row) and 55° (6th row) relative to the B₀ field, with TEs of 0 ms (I, i), 4.4 ms (J, j), 8.8 ms (K, k), and 13.2 ms (L, l). More dramatic signal decay is observed for the latter two sequences

Figure 4 shows exponential fitting curves for the same global ROI drawn in a patellar sample oriented 0°, 55°, and 90° to the B₀ field using 3D UTE Cones‐AdiabT_1ρ, Cones‐CW‐T_1ρ, and Cones‐ ${\mathrm{T}}_2^{\ast }$ imaging, respectively. The 3D UTE Cones‐AdiabT_1ρ value shows the smallest changes with angular orientations. The 3D UTE Cones‐CW‐T_1ρ value shows much increased change with angular orientations, while the 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$ value shows the most dramatic changes with angular orientations.

FIGURE 4.

Exponential fitting curves for a global ROI of a patellar sample oriented parallel (1st row), 55° (2st row), and 90° (3rd row) to the B₀ field using 3D UTE Cones‐AdiabT_1ρ imaging (A,D,G), regular 3D UTE Cones‐T_1ρ imaging (B,E,H), and 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$ imaging (C,F,I). AdiabT_1ρ values show the smallest magic angle effect, with 25.6% increase from 86.6 ms to 108.8 ms. Regular T_1ρ values show increased magic angle effect with, 76.5% increase from 27.2 ms to 48.0 ms. ${\mathrm{T}}_2^{\ast }$ values show the largest magic angle effect, with 205.0% increase from 12.1 ms to 36.9 ms

Figure 5 shows the magic angle behavior of 3D UTE Cones‐AdiabT_1ρ for all four ROIs (the superficial, transitional, and deep radial layers, and the global ROI comprised of all three layers) of a patellar sample oriented 0°, 55°, and 90° to the B₀ field, respectively. Single‐component exponential curve fitting was achieved for all layers at all angular orientations. The 3D UTE Cones‐AdiabT_1ρ value increases from the deep radial layer to the superficial layer for all angular orientations. Meanwhile, the superficial layer shows the least angular dependence, while the deep radial layer shows the strongest angular dependence.

FIGURE 5.

Representative 3D UTE Cones‐AdiabT_1ρ imaging of a patellar sample (A‐D) shows the three layer ROIs (30% deep radial, 60% transitional, and 10% superficial) and a global ROI (all three layers) used for fitting of AdiabT_1ρ at 0° relative to the B₀ field (1st row) for the deep radial (E), transitional (F), and superficial (G) layers of cartilage, and global ROI (H); at 55° relative to the B₀ field (2nd row) for the deep radial (I), transitional (J), and superficial (K) layers of cartilage, and global ROI (L); and at 90° relative to the B₀ field (3rd row) for the deep radial (M), transitional (N), and superficial (O) layers of cartilage, and global ROI (P). AdiabT_1ρ increases from the deep radial layer to the superficial layer for all angular orientations

Figure 6 shows the angular dependence of 3D UTE Cones‐AdiabT_1ρ, Cones‐CW‐T_1ρ, and Cones‐ ${\mathrm{T}}_2^{\ast }$ for all four ROIs (the superficial, transitional, and deep radial layers, and the global ROI comprised of all three layers) of a patellar sample. The 3D UTE Cones‐AdiabT_1ρ values show much reduced magic angle effect compared to both the regular 3D UTE Cones‐CW‐T_1ρ and Cones‐ ${\mathrm{T}}_2^{\ast }$ values for each layer individually, as well as the global ROI.

FIGURE 6.

3D UTE Cones‐AdiabT_1ρ profile (as a function of angular orientation) for the superficial (A), transitional (B), and deep radial (C) ROIs, as well as a global ROI (D). Regular 3D UTE Cones‐T_1ρ profile for the superficial (E), transitional (F), and deep radial (G) regions, as well as a global ROI (H). 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$ profile for the superficial (I), transitional (J), and deep radial (K) regions, as well as a global ROI (L). The superficial layers show reduced magic angle effect compared to the transitional and deep radial layers of articular cartilage. The AdiabT_1ρ values show much reduced magic angle effect compared to regular T_1ρ and ${\mathrm{T}}_2^{\ast }$ values

Table 1 summarizes the mean values and SDs of 3D UTE Cones‐AdiabT_1ρ, Cones‐CW‐T_1ρ, and Cones‐ ${\mathrm{T}}_2^{\ast }$ for individual layer ROIs and for the global ROI of eight human cadaveric patella cartilage specimens. Excellent inter‐observer agreement (ICC = 0.978‐0.996, P < .001) was achieved between the two radiologists in quantitative analysis. On average, the patellar samples measured over the whole cartilage thickness show 237.5% increase in 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$, 76.9% increase in Cones‐CW‐T_1ρ, and 27.2% increase in Cones‐AdiabT_1ρ when re‐oriented from 0° to 55° relative to the B₀ field. The magic angle effect is less dramatic for the superficial layer of patellar cartilage, with 87.9% increase in 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$, 11.3% in 3D UTE Cones‐CW‐T_1ρ, and 4.4% increase in 3D UTE Cones‐AdiabT_1ρ. The transitional layer shows increased angular dependence, with 262.9% increase in 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$, 59.1% in 3D UTE Cones‐CW‐T_1ρ, and 23.8% increase in 3D UTE Cones‐AdiabT_1ρ. The deep radial layer shows the most dramatic angular dependence, with 327.3% increase in 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$, 117.8% in UTE Cones‐CW‐T_1ρ, and 34.4% increase in 3D UTE Cones‐AdiabT_1ρ. Overall, the 3D UTE Cones‐AdiabT_1ρ biomarker shows greatly reduced magic angle effect, whether for different individual layers or for a global ROI comprising all layers of the articular cartilage.

TABLE 1.

3D UTE Cones‐AdiabT_1ρ, Regular 3D UTE Cones‐T_1ρ and 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$ mean values and SDs of eight normal patellar articular cartilage samples measured with the patellar cartilage oriented 0°, 30°, 55°, 75°, and 90° relative to the B₀ field

		Superficial (10%)	Transitional (60%)	Deep radial (30%)	Global ROI
Adiab T1ρ (ms)	0°	126.4 ± 6.9	100.2 ± 8.4	70.7 ± 14.6	92.9 ± 16.2
	30°	126.7 ± 6.0 (0.2%)	109.6 ± 10.4 (9.4%)	79.0 ± 9.2 (11.7%)	100.8 ± 18.2 (8.5%)
	55°	131.9 ± 10.2 (4.4%)	124.0 ± 9.3 (23.8%)	95.0 ± 7.5 (34.4%)	118.2 ± 10.6 (27.2%)
	75°	127.1 ± 12.5 (0.6%)	118.4 ± 7.9 (18.2%)	87.6 ± 7.8 (23.9%)	114.5 ± 11.5 (23.3%)
	90°	129.9 ± 18.4 (2.8%)	115.3 ± 12.3 (15.1%)	81.5 ± 7.8 (15.3%)	108.7 ± 13.0 (17.0%)
Cones‐T1ρ (ms)	0°	52.2 ± 4.6	32.0 ± 3.3	15.7 ± 3.0	26.4 ± 5.2
	30°	53.5 ± 9.3 (2.5%)	40.8 ± 6.9 (27.5%)	21.5 ± 2.0 (36.9%)	34.6 ± 7.9 (31.1%)
	55°	58.1 ± 7.4 (11.3%)	50.9 ± 8.6 (59.1%)	34.2 ± 3.1 (117.8%)	46.7 ± 7.9 (76.9%)
	75°	55.8 ± 9.4 (6.9%)	45.9 ± 5.4 (43.4%)	28.7 ± 3.2 (82.8%)	42.5 ± 7.7 (61.0%)
	90°	54.7 ± 6.5 (4.8%)	42.7 ± 9.5 (33.4%)	25.1 ± 3.9 (59.9%)	37.8 ± 8.0 (43.2%)
Cones‐ ${\mathrm{T}}_2^{\ast }$ (ms)	0°	24.7 ± 3.9	11.6 ± 5.4	5.5 ± 1.5	10.4 ± 4.1
	30°	30.0 ± 8.6 (21.5%)	16.6 ± 6.0 (43.1%)	8.7 ± 1.5 (58.2%)	15.4 ± 4.7 (48.1%)
	55°	46.4 ± 9.9 (87.9%)	42.1 ± 9.3 (262.9%)	23.5 ± 5.8 (327.3%)	35.1 ± 9.9 (237.5%)
	75°	31.8 ± 7.1 (28.7%)	28.7 ± 6.6 (147.4%)	15.5 ± 4.9 (181.8%)	26.5 ± 7.7 (154.8%)
	90°	29.5 ± 7.0 (19.4%)	22.9 ± 7.1 (97.4%)	11.6 ± 3.2 (110.9%)	19.3 ± 6.6 (85.6%)

The values in parentheses are percentage changes relative to the quantitative values at 0° relative to the B₀ field.

4. DISCUSSION

This is the first systematic study of the magic angle effect in 3D UTE Cones‐AdiabT_1ρ imaging of articular cartilage. We have demonstrated that magic angle‐less sensitive T_1ρ can be achieved for patellar cartilage using the 3D UTE Cones‐AdiabT_1ρ sequence. Furthermore, the angular dependence in 3D UTE Cones‐AdiabT_1ρ imaging was investigated not only for superficial and transitional layers of patellar cartilage, which is the focus of conventional T_1ρ and T2 imaging sequences, but for the deep radial layer, as well, which has a short T2 and is typically “invisible” with conventional sequences. The different layers of patellar cartilage have distinct 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$, Cones‐CW‐T_1ρ, and Cones‐AdiabT_1ρ relaxation times, which increase the deep radial cartilage to the superficial cartilage. Additionally, the different biomarkers exhibit very different magic angle effects, with the strongest magic angle effect occurring in the deep radial layer and the weakest magic angle effect in the superficial layer. The 3D UTE Cones‐AdiabT_1ρ relaxation time can be considered magic angle‐less sensitive compared to 3D UTE Cones‐ ${\mathrm{T}}_2^{\ast }$ and Cones‐CW‐T_1ρ biomarkers.

The 3D UTE Cones‐AdiabT_1ρ sequence has three major advantages over the conventional quantitative T2 and T_1ρ imaging sequences. First, the 3D UTE Cones‐AdiabT_1ρ sequence, similar to the regular AdiabT_1ρ sequence, ¹⁷ , ¹⁸ is expected to be less sensitive to the magic angle effect than conventional T2 and T_1ρ measured with a spin‐lock power of around 500 Hz. Although conventional T2 and T_1ρ were not measured in this study due to scan time limitations, our prior studies have demonstrated significant magic angle effects in both biomarkers, with a dramatic increase of 231.8% in CPMG‐T2 and 92% in magnetization‐prepared angle‐modulated partitioned‐k‐space spoiled gradient echo snapshots sequence for T1rho measurement (MAPSS T_1ρ) when samples were re‐oriented from 0° to 55° relative to the B₀ field. ¹⁰ Therefore, 3D UTE Cones‐AdiabT_1ρ measurement is expected to be reduced by 204.6% compared to CPMG‐T2, and reduced by 64.8% compared to MAPSS T_1ρ in terms of angle‐dependent changes. From this point of view, the 3D UTE Cones‐AdiabT_1ρ sequence likely provides an improved and more robust biomarker for cartilage degeneration.

Second, the 3D UTE Cones‐AdiabT_1ρ sequence allows T_1ρ mapping for all cartilage layers, including the superficial layer, the transitional layer, and the deep radial layer. This is a significant advantage over conventional T_1ρ and T2 sequences, which are only able to image and quantify the superficial and transitional layers of articular cartilage with their relatively long T2s. More recent studies suggest that OA is a systemic disease involving all major tissues and tissue components in the joint, including not only the superficial and transitional layers of articular cartilage, but also the deep radial cartilage, menisci, ligaments, tendons, and subchondral bone plate. ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ The 3D UTE Cones‐AdiabT_1ρ sequence allows systematic evaluation of all joint tissues and tissue components, and produces measurements which are less sensitive to the magic angle. Therefore, it is likely that this sequence will provide much improved diagnosis of OA, particularly in its early stages.

Third, the 3D UTE Cones‐AdiabT_1ρ sequence is based on a true 3D spiral acquisition with conical view ordering. The k‐space trajectory is more radial in the center of k‐space in order to achieve fast coverage of low spatial frequency data, and more curved in the outer k‐space in order to achieve higher sampling efficiency. ³² As a result, this sequence is highly time‐efficient, while providing motion‐insensitive quantitative imaging. ⁴¹ On the contrary, conventional T_1ρ and T2 imaging sequences are based on Cartesian encoding and are very sensitive to motion artifacts. The 3D UTE Cones‐AdiabT_1ρ sequence’s reduced motion sensitivity is another considerable benefit for potential clinical applications.

While our results are consistent with a series of recent studies on cartilage samples which have shown that conventional AdiabT_1ρ imaging is less sensitive to the magic angle effect, ¹⁶ , ¹⁷ , ¹⁸ it is not clear why the 3D UTE Cones‐AdiabT_1ρ sequence is more robust to angular dependence. One explanation might be that the adiabatic spin‐locking pulse has a relatively high power, which helps reduce angular dependence. CW‐T_1ρ imaging studies have demonstrated that the angular dependence is highly related to the power of the spin‐lock pulse ¹⁸ —in other words, that higher power leads to reduced magic angle effect. Another possible explanation is that the pairs of adiabatic spin‐locking pulses have a relatively broader spectral bandwidth than that of the CW spin‐locking pulse. Furthermore, one of the main advantages of using trains of adiabatic fast passage pulses for spin‐locking is the insensitivity to B1 inhomogeneity, which is a major limitation associated with conventional CW spin‐locking pulses. ¹³ More research is needed to explain the reduced magic angle sensitivity of the 3D UTE Cones‐AdiabT_1ρ sequence compared with the 3D UTE Cones‐CW‐T_1ρ sequence and conventional CW‐T_1ρ sequences.

There are several limitations of this study. First, the 3D UTE Cones‐AdiabT_1ρ sequence has been shown to provide magic angle‐less sensitive T_1ρ measurements for all cartilage layers. However, since histology was not performed in this study, it is not absolutely certain whether each patella was normal or degenerated (we determined this by visual inspection in our study). However, since the 3D UTE Cones‐AdiabT_1ρ biomarker showed little angular dependence for all patellar samples, it is likely that the magic angle effect would be reduced in either case of normal or degenerated cartilage. Second, it is still unclear whether this biomarker is sensitive to cartilage degeneration, although prior studies have demonstrated the sensitivity of conventional AdiabT_1ρ sequences to early degenerative changes in both animal and human specimens. ⁴² , ⁴³ , ⁴⁴ A sequential enzymatic digestion study of articular cartilage samples or a group of cadaveric specimens with different degrees of abnormality could help demonstrate the efficacy of this biomarker in detecting early cartilage degeneration, and will be conducted in the near future. Third, only patellar samples were investigated in this study even though a major advantage of the 3D UTE Cones‐AdiabT_1ρ sequence is its capability to systematically map T_1ρ of all main knee joint tissues or tissue components. It is unclear what the angular dependence of this biomarker is for other joint tissues, such as the menisci, ligaments, tendons, etc. Fourth, the clinical application of this biomarker is not investigated. Our prior studies have demonstrated the capability of this sequence for volumetric 3D UTE Cones‐AdiabT_1ρ mapping of the whole knee joint in vivo. ³⁶ A systematic study of healthy volunteers and patients with different degrees of OA is needed to demonstrate the clinical potential of this novel biomarker.

5. CONCLUSIONS

The 3D UTE Cones‐AdiabT_1ρ biomarker shows greatly reduced magic angle effect compared to regular 3D UTE Cones‐CW‐T_1ρ and Cones‐ ${\mathrm{T}}_2^{\ast }$. The 3D UTE Cones‐AdiabT_1ρ sequence may provide magic angle‐less sensitive evaluation of articular cartilage as well as other major knee joint tissues.

Wu M, Ma Y, Kasibhatla A, et al. Convincing evidence for magic angle less‐sensitive quantitative T_1ρ imaging of articular cartilage using the 3D ultrashort echo time cones adiabatic T_1ρ (3D UTE cones‐AdiabT_1ρ) sequence. Magn Reson Med. 2020;84:2551–2560. 10.1002/mrm.28317