Comparison of Normalization Methods for Selecting a Non-Small Cell Lung Cancer Marker Panel from Circulating miRNA RT-qPCR Data

The stability of miRNAs in biological fluids and their association with pathological conditions make miRNAs promising biomarkers for non-invasive disease diagnostics, including non-small cell lung cancer (NSCLC). However, variability in miRNA expression and technical aspects of quantitative reverse transcription PCR (RT-qPCR) necessitate effective normalization methods to ensure accurate evaluation of miRNA levels and identification of biological differences. In this study, we performed a comparative analysis of several miRNA normalization approaches applied to extracellular vesicles isolated from plasma of NSCLC patients, including pairwise normalization, “Tres” and “Quadro” normalization strategies, normalization to the mean, and normalization considering miRNA functional groups. Method effectiveness was evaluated using quality metrics of diagnostic models. The most robust results were observed with normalization methods utilizing miRNA pairs, triplets, and quadruplets, which provided high accuracy, model stability, and minimal overfitting. In contrast, normalization to the general or exclusive mean and functional group-based normalization showed lower efficiency in terms of classification performance and feature selection stability. These findings highlight the critical importance of normalization strategy choice to improve the accuracy and interpretability of RT-qPCR-based diagnostic models, particularly in the development of biomarker panels for NSCLC diagnostics. Pairwise normalization, combining computational simplicity and high efficiency, appears optimal for practical applications.