Analysis of liquid biopsy by Raman spectroscopy to facilitate prediction of response to immunotherapy in non-small-cell lung cancer (NSCLC) patients

Abstract

Immunotherapy has revolutionized lung cancer treatment, yet predicting patient response remains a challenge. This study used Raman spectroscopy to differentiate between non-small-cell lung cancer patients with short-lasting and long-lasting responses to the first line of immunotherapy. Notable spectral shifts were observed at 1343 cm^-1 and 2934 cm^-1, distinguishing short-lasting from long-lasting responders. A classification model achieved 75 % accuracy, with 81 % sensitivity and 67 % specificity in predicting response duration. The band at 1343 cm^-1 emerged as a potential Raman biomarker for long-lasting responses, offering a promising tool for treatment stratification and personalized therapy approaches. Importantly, the proposed band correlated with the number of white blood cells obtained from short-lasting responders, while in long-lasting responders, this band correlated with thyroid hormones concentration, red blood cells number, hemoglobin level, and lymphocytes number. The correlations between medical parameters and Raman data showed the reflectance of lung cancer patients' conditions in the Raman spectra of plasma.

Keywords: Immunotherapy; Liquid biopsy; Lung cancer; Machine learning; NSCLC; Raman spectroscopy.