P07.12.A DIAGNOSTIC POTENTIAL OF INFRARED SPECTROSCOPY IN MENINGIOMA PATIENTS

Abstract

Abstract BACKGROUND Vibrational spectroscopy probes molecular vibrations in the tissue and can provide optical diagnosis of brain tumors. Here, we investigated its potential for the identification of aggressive meningioma. MATERIAL AND METHODS Tissue samples were obtained during routine surgery from 113 meningioma patients (WHO 1 n= 33; WHO 2 n=39; WHO 3 n=41). Infrared (IR) spectroscopic imaging was performed at three positions (170 x 170 µm, 16x16 = 256 spectra) on frozen sections in reference to HE histology. Spectra were labeled according to the clinical diagnosis of the patient. Baseline correction, area normalization and fourfold spectral binning was performed. Spectra displaying artifacts were identified by visual inspection and were excluded from further analysis. The data was split into test and training set. 42825 spectra of 57 specimens (training set) were utilized to develop an algorithm using linear discriminant analysis to identify aggressive WHO 3 meningioma. RESULTS The developed algorithm assigned 88.6% of the 40626 spectra of the independent test set to the correct class. Based on these RESULTS , the class assignment for each measurement position was calculated as mean of class assignment of the respective spectra: All WHO 1 (16/16) samples, 16/19 WHO 2 and 17/21 WHO 3 meningioma were correctly assigned at all three positions. One sample of meningioma WHO 3 was misclassified at all measurement positions. However, in case of six patients inconclusive results were obtained for the samples as the measurement positions were assigned to different classes. Comparison of IR spectral signatures revealed increased bands at 1114, 1242, 1336 and 1450 cm-1 and decreased bands at 1544 and 1651 and 1734 cm-1 in meningioma WHO 3. This suggests increased phospholipids and decreased proteins as main difference to meningioma WHO 1 and 2 and might form the basis for diagnostic exploitation of IR spectral information. CONCLUSION IR spectroscopy qualifies as reliable tool for the identification of aggressive meningioma. Based on the classification results of this study, translational strategies may be deduced. The data suggests, that analysis of the tumor needs to be performed at multiple sites to obtain reliable clinical information. In case of conflicting results at different measurement positions, the tissue should be evaluated as unratable. As infrared spectroscopy can be applied in situ using fiber based systems, it might allow intraoperative meningioma grading and tumor delineation in future translational settings.