Dataset Information

Machine learning-based prediction of glioma margin from 5-ALA induced PpIX fluorescence spectroscopy.

ABSTRACT: Gliomas are infiltrative brain tumors with a margin difficult to identify. 5-ALA induced PpIX fluorescence measurements are a clinical standard, but expert-based classification models still lack sensitivity and specificity. Here a fully automatic clustering method is proposed to discriminate glioma margin. This is obtained from spectroscopic fluorescent measurements acquired with a recently introduced intraoperative set up. We describe a data-driven selection of best spectral features and show how this improves results of margin prediction from healthy tissue by comparison with the standard biomarker-based prediction. This pilot study based on 10 patients and 50 samples shows promising results with a best performance of 77% of accuracy in healthy tissue prediction from margin tissue.

SUBMITTER: Leclerc P

PROVIDER: S-EPMC6989497 | biostudies-literature | 2020 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine learning-based prediction of glioma margin from 5-ALA induced PpIX fluorescence spectroscopy.

Leclerc Pierre P Ray Cedric C Mahieu-Williame Laurent L Alston Laure L Frindel Carole C Brevet Pierre-François PF Meyronet David D Guyotat Jacques J Montcel Bruno B Rousseau David D

Scientific reports 20200129 1

Gliomas are infiltrative brain tumors with a margin difficult to identify. 5-ALA induced PpIX fluorescence measurements are a clinical standard, but expert-based classification models still lack sensitivity and specificity. Here a fully automatic clustering method is proposed to discriminate glioma margin. This is obtained from spectroscopic fluorescent measurements acquired with a recently introduced intraoperative set up. We describe a data-driven selection of best spectral features and show h ...[more]

PMID: 31996727

Similar Datasets

Project description:5-Aminolevulinic acid (5-ALA) induced fluorescence to augment surgical resection for high grade glioma has become a standard of care. Protoporphyrin IX (PpIX) visibility is however subject to the variability of the single tumor expression and to the interobserver interpretation. We therefore hypothesized that in different glioma cell lines with variable 5-ALA induced fluorescence, the signal can be pharmacologically increased. We therefore analyzed in three different GBM cell lines, with different expression of epidermal growth factor receptor (EGFR), the variability of 5-ALA induced PpIX fluorescence after the pharmacological blockade at different steps of PpIX breakdown and influencing the outbound transport of PpIX. Using flow cytometry, fluorescence microplate reader, and confocal microscopy the PpIX fluorescence was analyzed after exposure to tin protoporphyrin IX (SnPP), deferoxamine (DFO), and genistein. We furthermore constructed a microscope (Qp9-microscope) being able to measure quantitatively the concentration of PpIX. These values were compared with the extraction of PpIX in tumor biopsy taken during the GBM surgery. Although all three cell lines showed an increase to 5-ALA induced fluorescence their baseline activity was different. Treatment with either SnPP, DFO and genistein was able to increase 5-ALA induced fluorescence. Qp9-microscopy of tumor sample produced a color coded PpIX concentration map which was overlaid on the tumor image. The PpIX extraction from tumor sample analyzed using the plate reader gave lower values of the concentration, as compared to the expected values of the Qp9-microscope, however still in the same decimal range of ?g/mL. This may be due to homogenization of the values during extraction and cell disaggregation. In conclusion pharmacological augmentation in GBM cell lines of PpIX signal is possible. A quantitative PpIX map for surgery is feasible and may help refine surgical excision. Further correlations of tumor tissue samples and Qp9-microscopy is needed, prior to develop an intraoperative surgical adjunct to the already existing 5-ALA induced surgery.

Project description:The extent of 5-aminolevulinic acid (5-ALA) guided tumor resection has a determining impact in high-grade glioma and glioblastoma surgery. Yet the intensity of the 5-ALA induced fluorescence may vary within the tumor. We aimed to correlate 5-ALA induced fluorescence with the expression of epithelial growth factor receptor (EGFR) and its constitutively active version EGFRvIII in different glioblastoma (GBM) cell lines. To elucidate the role of EGFR in the metabolism of 5-ALA in GBM cell lines with variable EGFR expression status, we analyzed the activation of EGFR by its primary ligand EGF, and its downstream effect on Heme oxygenase-1 (HO-1), a key enzyme regulating the metabolism of Protoporphyrin IX (PpIX), the fluorescent metabolite of 5-ALA. Effects of direct pharmacological inhibition by Tin(IV)-Protoporphyrin (SnPP) or gene knockdown by small interfering RNA (siRNA) on HO-1 enzyme were analyzed in respect to 5-ALA induced fluorescence. Furthermore, inhibition of EGFR by Gefitinib was tested. A significant difference in 5-ALA induced fluorescence was obtained in U87MG (low EGFR expression) and LN229EGFR cells (EGFR overexpression) compared to BS153 (EGFR overexpression/EGFRvIII+). Treatment of U87MG and LN229EGFR cells with EGF significantly reduced cellular fluorescence, by promoting HO-1 transcription and expression in a concentration-dependent manner. This effect could be reversed by EGFR-specific siRNA treatment, which reduced protein expression of about 80% in U87MG. Remarkably, inhibition of HO-1 activity by SnPP or reduction of HO-1 protein levels by siHO-1 treatment restored fluorescence in all cell lines, independently of EGFR quantitative and qualitative expression. Gefitinib treatment was able to restore fluorescence after EGF stimulation in U87MG cells but not in BS153 cells, overexpressing EGFR/EGFRvIII. In GBM cell lines, 5-ALA induced fluorescence is variable and influenced by EGF-induced downstream activation of HO-1. HO-1 protein expression was identified as a negative regulator of 5-ALA induced fluorescence in GBM cells. We further propose that co-expression of EGFRvIII but not quantitative EGFR expression influence HO-1 activity and therefore cellular fluorescence.

Dataset Information

Machine learning-based prediction of glioma margin from 5-ALA induced PpIX fluorescence spectroscopy.

Publications

Machine learning-based prediction of glioma margin from 5-ALA induced PpIX fluorescence spectroscopy.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets