Fluorescence Spectroscopy for Quantitative Demarcation of Glioblastoma Using 5-Aminolevulinic Acid

Sammanfattning: Total resection of glioblastoma, the highly malignant brain tumor, is difficult to accomplish due to its diffuse growth and similarity to the surrounding brain tissue. A total resection is proven to increase patient survival. The aim of this thesis was to evaluate fiber-optical based fluorescence spectroscopy for quantitative demarcation of malignant brain tumors during the surgery. Five-aminolevulinic acid (5-ALA) was used as a fluorescence contrast agent that accumulated as protoporphyrin IX (PpIX) in the tumor.The method was evaluated at the Department of Neurosurgery, Linköping University Hospital. The patients (n = 22) received an oral dose of 5 mg/kg body weight 5-ALA two hours prior to craniotomy. Measurements with a developed fluorescence spectroscopy system were performed under the general procedure of surgery. The collected fluorescence spectra were quantified by defining a fluorescence ratio and the main challenges of measuring and quantifying spectra were investigated. The fluorescence ratio was compared to visual diagnosis of the surgeon, histopathological examination and ultrasound-based neuronavigation. The main challenges of using a fluorescence spectroscopy system in the operating room were the disturbing ambient light, photobleaching and blood interference which affect the signal quantification. The superimposition of ambient light was removed by modulating the system.Using principal component analysis (PCA) the photobleaching sequences could be described by three spectral components of autofluorescence, PpIX fluorescence and blue-shift. To investigate the photobleaching induced prior to the measurements, a dynamic model was developed based on the PCA derived spectral components. Modulation and increased power of the excitation light resulted in a faster photobleaching; however, photobleaching was saturated at higher excitation powers. The system was adjusted to induce minimal photobleaching. In addition, effect of blood absorption on the fluorescence spectrum was investigated experimentally by placing blood drops on skin and theoretically by using Beer-Lambert law. The theoretical model was used to compensate for the distorted fluorescence ratio. According to the theoretical model of blood interference, a total 300 µm blood layer blocked the brain fluorescence signal totally and when the fluorescence signal was partially blocked, the fluorescence ratio was overestimated. The fluorescence ratio was corrected for blood layers thinner than 50 µm.The tissue in and around the tumor was categorized into necrosis, low and high grade tumor and gliosis. The median fluorescence ratio confirmed with histopathological examination (n = 45) had a lower fluorescence ratio for low grade malignancies (0.3) than high grade malignancies (2.4) (p < 0.05). Gliosis (1.6) and necrosis (1.0) showed a moderate fluorescence ratio. Ultrasound-based navigation in combination with fluorescence spectroscopy showed improvement in the results; however, a more extensive study is needed to confirm benefits of the method combination. In conclusion, fluorescence spectroscopy of 5-ALA induced PpIX provided an objective method for differentiating tumor from the healthy tissue intra-operatively. Fluorescence ratios were indicative of tissue type and tumor malignancy degree.