Introduction

Loss of collagen is associated with cartilage degeneration and loss of Glycosaminogycans (GAGs) are hallmarkers of early osteoarthritis^1-3. Previously, only few studies have focused on simulating the effects of chemical exchange between free water and GAGs exchanging groups to differentiate degenerated cartilage from intact tissue. Relaxation along a Fictitious Field (RAFF) provides a flexible approach to adjust pulse parameters, especially the pulse duration, to sensitize measurements for chemical exchange. This is achieved at an acceptable specific absorption rate and scanning time. The purpose of this study was to study the RAFF pulse duration effects on proton exchange between free water and -OH groups and relaxation time difference between intact and degraded cartilage.

Methods

Simulations were performed assuming a two-pool model of proton exchange between free water and GAG hydroxyl group (OH) using Bloch McConnell equations^2,3. An exchange rate of 1000 Hz between water and GAG-OH, concentration 0.3 M of GAG, T₁ = 1 s, T₂ = 0.01 s for both pools and chemical shift 1 ppm between water and GAG-OH were assumed^1-3.
Twelve samples with collagen concentration 20, 40, and 60 mg/mL, and chondroitin sulphate concentration of 0, 10, 20 and 40 mg/mL were mixed to create semisolid collagen phantoms. Two osteochondral specimens were collected from two patients with idiopathic knee OA during knee joint arthroplasty. The specimens are from the central load-bearing region of the lateral femoral condyle (the right sample in Fig. 2) and the other from the posterior aspect of the medial femoral condyle. Both specimens presented both intact and degenerated cartilage areas.
RAFF pulse waveform was created with initial 500 Hz amplitude and embedded in the refocusing scheme leading to a pulse duration (T_p) 2.83 ms^4,5. For RAFF relaxation time (T_RAFF) mapping, pulses were repeated [0,12,24,36] times to form the RAFF pulse trains. A crusher gradient was applied to dephase the magnetization in xy-plane prior to readout. The RAFF imaging parameters were as follows: gradient echo readout with repetition time (TR) 6 s between pulse trains, echo time (TE) 5.9 ms, matrix size 192×156, flip angle 20°, field of view 100 × 69 mm², and slice thickness 3 mm. Experiments were conducted at 3 Tesla.
A T₂-prepared multi-echo spin-echo sequence was used to acquire five T₂-weighted images (preparation TE = 12, 23, 35, 47, 59 ms). T₂ maps were generated using a log-transformed linear least-squares fitting. The T₂ mapping readout parameters were TR=2000 ms, field of view 100 × 69 mm², flip angle 70°, and slice thickness 2 mm. A factorial ANOVA was conducted to compare the main effect of collagen on RAFF relaxation times for both phantoms and specimens. Due to limited number of samples, each pixel of region of interests was considered as independent point in statistical analysis. In addition, the estimated marginal means of relaxation times and partial Eta squared were extracted from the factorial analysis for phantoms. Correlation coefficients of Spearman’s rho and corresponding P-values were calculated to investigate the relations between collagen phantoms and simulations. Differences between degraded and intact cartilage with RAFFs with varying pulse durations were tested with t-test and corrected for multiple comparisons using Benjamin-Hockert test.

Results

Although collagen concentrations were low in phantoms compared to cartilage, significant differences in T_RAFF curves with pulse durations 1.8-3.4 ms were found between collagen concentrations (***p<0.001) (Fig 1a). Simulations demonstrated a decreased relaxation times with increasing concentration of -OH group (Fig 1d). Significant Spearman’s correlation coefficients were observed between collagen concentrations and corresponding simulations, (Fig 1a and 1d) (R² = 1, **p<0.01 for collagen 20 and 40 and R² = 0.943, **p<0.01 for collagen 60). The highest contrast was obtained at pulse duration of 2.8 ms (collagen intergroup effect Partial eta² = 0.603, ***p<0.001) (Fig 1c). Significant difference between degraded and intact cartilage observed only in RAFF with pulse duration of 2.8 ms (t-test p=0.04) and T2 (t-test p=0.014) (Fig. 2 and 3a). The normalized T_RAFF exhibited significant increase in degraded compared to intact cartilage. Considering normalized specimens’ measurements, pairwise comparison of factorial ANOVA indicates significant difference between degraded and intact cartilage only in T₂ map and T_RAFF with pulse duration 2.8 ms (Mean Diff = 0.16, ***p<0.001 and Mean Diff = 0.10, *p=0.034) (Fig. 3b).

Discussion

The results from cartilage-mimicking phantoms, simulations and specimens demonstrated that increasing pulse duration of RAFF relaxation time until ~2.8 ms raise the sensitivity to loss of -OH group in degraded cartilage independently on -OH group origin (collagen or proteoglycan). Progressed degeneration specimen’s most likely reflect collagen loss which is consistent with the phantom measurements. The T_RAFF increases in degraded cartilage compared to intact which is in line with decreased collagen and/or GAG-OH concentration. The limitation of the study was the low collagen concentration in phantoms compared to cartilage and a low signal-to-noise ratio in specimen measurements. We tackled the problem by normalizing specimens’ measurements, particularly visible at lowest pulse durations in RAFF measurement.

Conclusion

We have demonstrated that RAFF pulse duration near to 2.8 ms is optimal to detect cartilage degeneration with RAFF.

Acknowledgements

The authors want to thank financial support from following organizations Academy of Finland, Finnish Foundation for Carciovascular Research and Aarne Koskelo Foundation.