Detection of radiosensitive subpopulations ex-vivo with Raman microspectroscopy

Area of Science:

• Biomedical Spectroscopy
• Cancer Research
• Radiobiology

Background:

• Understanding radiosensitivity is crucial for effective cancer therapy.
• Current methods for detecting radiosensitivity are limited.
• Raman spectroscopy has emerged as a promising tool for biological analysis.

Purpose of the Study:

• To investigate Raman spectroscopy as a method for detecting and classifying radiosensitivity.
• To evaluate the performance of machine learning models in discriminating radiosensitive cell lines.
• To explore the potential of Raman spectroscopy for liquid biopsy applications.

Main Methods:

• Raman spectroscopy was used to analyze lymphoblastoid cell lines from patients with ataxia telangiectasia, non-Hodgkins lymphoma, Turner's syndrome, healthy controls, and prostate cancer patients.
• Cell samples were exposed to varying doses of X-ray irradiation (0 Gy, 50 mGy, 500 mGy).
• Support vector machine models, utilizing linear (PCA) and non-linear (UMAP) dimensionality reduction, were developed to classify spectral data.

Main Results:

• Raman spectroscopy, combined with machine learning, successfully discriminated between radiosensitive and non-radiosensitive cell populations.
• Non-linear dimensionality reduction (UMAP) improved discrimination compared to linear methods (PCA).
• Models trained on non-irradiated samples (0 Gy) achieved the highest performance (F1 = 0.92 ± 0.06) in classifying cell types.

Conclusions:

• Raman spectroscopy holds potential as a non-invasive tool for assessing intrinsic radiosensitivity.
• The technique may enable the identification of radiosensitive subpopulations for personalized medicine.
• Further research could validate Raman spectroscopy for clinical use in liquid biopsies.