In vivo knee cartilage quality assessment by direct quantification of glycosaminoglycans through chemical exchange saturation transfer (gagCEST)

Introduction: Non-invasive imaging of human articular cartilage is an important tool in understanding, and thus allowing the development of effective treatments for, osteoarthritis; a disease responsible of impaired movement of millions of otherwise healthy individuals. Glycosaminoglycans (GAG) have been attributed with important biomechanical properties, vital for the function of the articular cartilage. By selectively irradiating hydroxyl protons at the GAG molecule and studying the saturation effect as it is transferred to water through chemical exchange, the GAG concentration can be inferred. This is termed glycosaminoglycan measurement through chemical exchange saturation transfer – gagCEST. This method would ideally directly measure the concentration of GAG in articular cartilage and increase the knowledge about osteoarthritis pathology. The aim of this thesis is to give the reader an understanding of the gagCEST method, and to evaluate a specific gagCEST work-in-progress (WIP) package. Materials and methods: A WIP package for gagCEST measurements was evaluated on a Magnetom Trio 3 Tesla MRI system (Siemens AG, Erlangen, Germany), compatible with a 15 channel transmit/receive knee-coil. Phantoms with variable amount of GAG and agarose were measured in order to investigate the function of the system. Freeze dried samples of human femoral cartilage were supplied by the orthopedics department in order to optimize the measurement. Finally, several volunteers were scanned to test the in vivo applicability of the sequence at its current stage. Evaluation of measurements were done both using the raw data to create a CEST-spectrum in MatLab, as well as relying on the pixel-wise calculation of the CEST-image produced by the WIP programme. Results: The gagCEST measurements proved to be filled with pitfalls. Phantom measurements were complicated by magnetic field inhomogeneity and incomplete gradient spoiling, grossly disturbing the CEST spectra. Though heterogeneity correction is applied, phantoms with lower T2 relaxation must be produced to get rid of residual magnetization after spoiling. Despite this, some measurements were performed yielding a noticeable CEST effect when considering mean value of several pixels. In vivo measurements were first disturbed by fat chemical shift, but proved to produce images with CEST signal in cartilage. However, the evaluation of in vivo images was difficult, e.g., the CEST signal ratio between cartilage and meniscus was one order of magnitude away from the theoretically expected result. Discussion: The gagCEST WIP sequence proved to provide image volumes in six minutes, displaying high signal in cartilage on healthy volunteers. However, the method requires additional extensive investigation. Phantom measurements are required to create standards and optimize the saturation sequence, for which low T2 and long T1 phantoms must be constructed. Moreover, in vivo measurements must be further scrutinized as magnetic field heterogeneity will quickly ruin a measurement; the cartilage of interest is tightly packed between bone and the meniscus. It is also important to realize that the gagCEST method does not in fact measure GAG, but rather labile hydroxyl protons, and there exist a possibility that other molecules will exhibit similar chemical shifts. Though this is not a physical but biochemical consideration it should be kept in mind when analyzing tissue and “gagCEST” signal. Finally, several authors claim the technique will be feasible at 7 T due to longer T1 and larger chemical shift of GAG hydroxyls, however, it is evident from phantom tests that at least relative measurements are possible at 3 T