A Novel Strategy for Targeted Lipidomics Based on LC-Tandem-MS Parameters Prediction, Quantification, and Multiple Statistical Data Mining: Evaluation of Lysophosphatidylcholines as Potential Cancer Biomarkers

TLDR

Lipid quantification is the ultimate goal in lipidomics, yet it is limited by the scarcity of standard compounds. This study develops a novel targeted lipidomics strategy that predicts LC‑MS/MS parameters and applies multivariate statistics to quantify lysophosphatidylcholines. The authors built multiple‑linear regression models linking acyl‑chain length and unsaturation to predict collision energy, declustering potential, retention time, and response factor, and integrated logistic‑regression ROC analysis with t‑test, ROC, and PLS‑DA to identify and evaluate LPC biomarkers. Using this model‑prediction workflow, sensitivity, accuracy, and coverage were markedly improved, enabling simultaneous determination of 60 LPCs in plasma and the identification of multi‑LPC index panels that discriminate lung, breast, colorectal, and gastric cancers from controls and among each other, suggesting clinical utility.

Abstract

Lipid quantification is the ultimate goal in lipidomics studies challenged by the availability of standard compounds. A novel strategy for targeted lipidomics based on LC-MS/MS parameters prediction and multivariate statistical analysis was developed for the quantitation of lysophosphatidylcholines (LPCs) in this study. Multiple linear regression models were established with the acyl chain length and number of double bonds after the prediction correlation coefficients (R2pred) were evaluated. Then related analytical parameters including collision energy, declustering potential, retention time, and response factor were successfully predicted for any given LPC. With this "model-prediction" strategy, sensitivity, accuracy, and coverage of targeted lipidomics were improved significantly, and 60 LPCs were determined simultaneously in plasma for the first time. An integrated evaluation method for multi-indexes, logistic regression-ROC analysis was also proposed after biomarkers were identified by Student's t test, univariate ROC curve, and PLS-DA. Then the developed workflow was successfully used to discover and evaluate multi-LPCs indexes (a set of LPCs biomarkers with the best discriminating ability) for differentiating lung, breast, colorectal, and gastric cancer from controls, and among different types of cancer. Finally, the multi-LPCs index for lung cancer was compared with the plasma before and after treatment to test its utility. The novel targeted lipidomics methodology for LPCs was expected to provide a new insight into quantitative lipidomics and further clinical application.