SERS-Based Serum Molecular Spectral Screening for Hematogenous Metastasis

Overview

Although modern medicine has made significant progress in the diagnosis and treatment of lung cancer, most patients are diagnosed at locally advanced stage or with distant metastases, especially in the late stages where the cancer has spread to other organs through hematogenous metastasis. This not only significantly the survival rate of patients but also increases the complexity and difficulty of treatment. Hematogenous metastasis plays an important role in the clinical progression of lung cancer, its complex biological processes pose a huge challenge for clinical management. Early detection of hematogenous metastasis is difficult, and traditional imaging methods have limited sensitivity in detecting small metastatic lesions. The emerging technology of circulating tumor cells (CTCs) has been limited in clinical application due to its high detection costs and technical requirements. Therefore researching and developing high-sensitivity, high-specificity, simple, easy-to-popularize, and low-cost technologies to predict the risk of hematogenous metastasis lung cancer is crucial for early diagnosis and more precise treatment. Raman spectroscopy (RS), a non-invasive and highly specific molecular detection technology, can detect in biomolecules such as proteins, nucleic acids, lipids, and sugars related to tumor metabolism in biological samples at the molecular level. Surface-enhanced R spectroscopy (SERS), developed based on this technology, is one of the feasible methods for high-sensitivity biomolecular analysis. Although SERS technology has shown diagnostic results in numerous preclinical studies of various tumors, it is limited by small sample sizes and lacks external validation. Therefore, clinical studies on the diagnosis of tumors Raman spectroscopy are needed, with the following requirements: 1. Objective, rapid, and practical Raman spectroscopy data processing methods are needed, and and deep learning methods may be the best classification methods; 2. Multicenter, large-sample clinical samples are needed to train deep learning diagnostic models, and real-world performance should be validated through external data from prospective studies. In previous study, the investigators collected serum Raman spectroscopy data from a cohort of 23 patients with lung malignancies and developed an intelligent Raman diagnostic system for hematogenous metastasis in non-small cell lung cancer (NSCLC) based on learning models, with an accuracy rate of 95%. To obtain the highest level of clinical evidence and truly achieve clinical translation, this prospective, multicenter clinical aims to validate the use of this intelligent diagnostic system for early diagnosis of hematogenous metastasis in NSCLC.

This study used a confocal Raman microspectrometer produced by Renishaw, Britain, purchased by the Key Laboratory of the School of Optoelectronics and Engineering of Fujian Normal University. The spectral resolution was 2 cm-1, the excitation wavelength was 785 nm, and a 20x objective Leica microscope was used to collect SERS spectra in the range of 400-1800 cm-1. The excitation irradiation time of each spectrum was 1 s, and the laser power was 30 mW. The measured SERS spectra were collected using the WIRE3.4 (Renishaw) software package. In order to reduce the interference of fluorescence background signals between different spectral lines, the Vancouver Raman Algorithm software (multi-order polynomial fitting algorithm) was used to remove the fluorescence background, remove the baseline and smooth the results. At the same time, in order to avoid changes in peak spectrum intensity caused by instrument performance problems, the spectrum after background subtraction was normalized using NILabVIEW2014 software. Then, the obtained spectral data was analyzed for mean spectrum and charts using Origin, and multivariate statistical analysis was performed.