SERS and Machine Learning-Enabled Liquid Biopsy: A Promising Tool for Early Detection and Recurrence Prediction in...

Acute leukemia (AL), classified as acute myeloid leukemia (AML) and acute lymphocytic leukemia (ALL), is a hematologic malignancy caused by the uncontrolled proliferation of leucocytes in the bone marrow. Early detection of AL is crucial for clinical treatment. Detection methods of AL are currently blood tests, bone marrow tests, imaging, and spinal fluid tests. However, these tests have drawbacks, such as high cost and time consumption. Liquid biopsy using biological fluids such as blood or serum is an emerging technique for noninvasive cancer detection and monitoring. Surface-enhanced Raman spectroscopy (SERS), which enhanced Raman signals by the interaction of plasmonic nanostructures with the analyte, is a highly sensitive and specific detection method with simple sample preparation that has been used in combination with machine learning techniques to analyze liquid biopsy. In this study, we developed a SERS-based liquid biopsy approach that enables accurate classification of AML and ALL subtypes and the prediction of disease recurrence. SERS spectra of serum samples from 24 healthy individuals, 43 AML patients, and 18 ALL patients were obtained using an Ag-based SERS substrate and clustered using hierarchical cluster analysis (HCA). The spectra were then classified using three commonly used classifiers, namely, support vector machine (SVM), random forest (RF), and k-nearest neighbor (kNN). Our findings demonstrate that the RF classifier has the highest accuracy values, with 96.1, 95.5, and 98.5% for classifying three groups and predicting the recurrence of AML and ALL, respectively. The combination of SERS-based serum analysis with machine learning algorithms represents a remarkable advancement in the realm of hematological disease diagnostics, particularly for AML and ALL. This approach not only facilitates the precise differentiation of disease subtypes but also introduces the novel capability of prognosticating disease recurrence.