Rapid genetic classification of gliomas using Raman spectroscopy

Background:

Raman spectroscopy probes the unique molecular vibrations of a sample to accurately characterise its molecular composition. No sample processing is required allowing for rapid analysis of fresh tissue. The genetic classification of gliomas, particularly isocitrate dehydrogenase (IDH) mutations, is critical for clinical decision-making. With the development of new drugs targeting specific glioma genetic subtypes it will become increasingly important for surgeons to be aware of the genetics of the tumour at the time of operation to inform their surgical strategy. The aim of this study was to use Raman spectroscopy on fresh samples taken straight from the operating theatre and to classify gliomas according to their genetic subtypes. Similar classification models were built using cryosections and formalin-fix paraffin embedded (FFPE) sections.

Methods:

Raman spectra were collected using a Renishaw benchtop RA800 series spectrometer. Parallel sections underwent immunohistochemistry with targeted genetic sequencing when required to confirm the following five glioma subtypes: Glioblastoma, IDH- mutated; glioblastoma, IDH-wildtype; astrocytoma, IDH-mutated; astrocytoma, IDH-wildtype; oligodendroglioma).

Results:

Fresh tissue samples from 50 patients were collected (6 glioblastoma, IDH- mutated; 33 glioblastoma, IDH-wildtype; 6 astrocytoma, IDH-mutated; 1 astrocytoma, IDH- mutated; 3 oligodendroglioma, 1 excluded). A principal component-linear discriminant analysis (PCA-LDA) model demonstrated 80%-95% sensitivity and specificity for predicting the five glioma genetic subtypes. For prediction of IDH mutation alone the model gave a 92% sensitivity and 91% specificity. 86 cryosection and 117 FFPE samples underwent Raman with models demonstrating 87-94% sensitivity and 77-80% specificity for predicting IDH mutations. In the fresh tissue samples, the mean time for spectra collection was 9.5 minutes with the whole process from tumour biopsy to genetic classification taking under 30 minutes.

Conclusion:

These results demonstrate proof of concept that Raman spectroscopy can be used for rapid, intraoperative glioma genetic classification. Further work is being done to refine model building to increase classification performance.