Lipids are crucial in metabolic and physiological processes and various diseases. While traditional clinical parameters like cholesterol and triglycerides are used, the clinical application of advanced lipidomics remains limited despite the growing evidence linking lipids to various diseases. Mass spectrometry (MS)-based lipidomics, particularly using technologies like ion mobility spectrometry (IMS), offers potential for high-throughput clinical profiling. Trapped ion mobility spectrometry (TIMS) coupled with parallel accumulation serial fragmentation (PASEF) provides four-dimensional (4D) lipidomics (m/z, retention time, collision cross-section (CCS), and fragmentation spectra), allowing for detailed lipid analysis. However, clinical translation requires improvements in accuracy, precision, robustness, and throughput. This study aims to develop and validate a high-throughput 4D TIMS lipidomics platform for comprehensive and reproducible clinical lipid profiling, addressing challenges in confident lipid annotation and data curation inherent in high-throughput applications.

Existing literature highlights the importance of lipids as disease biomarkers and therapeutic targets. Advanced MS techniques, including IMS, have improved lipid profiling capabilities. TIMS-PASEF offers 4D lipidomics, but faces challenges in annotation accuracy and standardization. Studies have shown high false discovery rates in lipid annotation if stringent parameters aren't used. The use of risk scores based on specific lipids, like ceramides and phosphatidylcholines (CERT1 and CERT2 scores), shows promise for cardiovascular disease risk assessment. However, widespread clinical translation requires improvements in method validation, standardization, and cross-laboratory validation. This research builds upon existing literature by focusing on developing a robust, high-throughput, and highly accurate 4D lipidomics workflow specifically suitable for clinical settings.

The study developed a high-throughput workflow combining automated lipid extraction with UHPLC-TIMS-PASEF-MS. A robotic platform adapted an existing liquid-liquid extraction (LLE) method for plasma lipids, improving throughput and reproducibility. Methyl tert-butyl ether (MTBE) was chosen for extraction due to its efficiency. Robotic parameters were optimized for efficient solvent removal. The recovery and matrix effects of internal standards (ISTDs) for various lipid classes were evaluated, demonstrating comparable performance to manual extraction and acceptable coefficients of variation (CVs). Microflow UHPLC provided high ionization efficiency and chromatographic resolution in under 20 minutes. An in-house 4D lipid library was created using 200 lipid standards, encompassing CCS, RT, m/z, and MS/MS spectra. This library was expanded by adding manually curated lipid species from NIST plasma SRM using the LipidIMMS analyzer, MS Dial, RT and CCS values. Stringent feature selection criteria were implemented based on RT, accurate mass, CCS, isotopic pattern, and MS/MS spectral matching. Two sets of ISTDs were used for quantification: a set of level-3 class-specific ISTDs and a set of deuterated ISTDs. Both multi-point and one-point quantification strategies were evaluated using TOF MS survey data. The method's performance was validated using NIST human plasma and serum standard reference materials (SRMs), with cross-validation by MRM. Finally, the workflow was applied to a pilot study (LBlooD) profiling intra-individual lipidome phenotypes across five blood matrices (plasma, serum, whole blood, venous DBS, and finger-prick DBS) from four individuals over three time points. Statistical analysis included Wilcoxon signed-rank tests, Friedman tests, random forest classification, and PCA to compare lipidome profiles.

The automated high-throughput lipid extraction showed comparable lipid recoveries and matrix effects compared to manual extraction, with most lipid species exhibiting CVs below 10%. Microflow UHPLC-TIMS-MS achieved high ionization efficiency and chromatographic resolution, partially resolving isomers. The in-house 4D lipid library significantly improved annotation accuracy and reduced false-positive assignments. High inter-day reproducibility was demonstrated for 4D feature detection (median variability of 0.58 ppm, median CVs of 0.19% RT and 0.11% CCS). 370 lipids were confidently annotated in NIST plasma SRM, and 364 in serum SRM. 359 lipids were reproducibly quantified in NIST plasma SRM and 354 in serum SRM. The one-point quantification strategy showed comparable results to multi-point quantification for many lipid classes. The CERT2 score, a predictive marker for cardiovascular disease, showed consistent results across assays. The LBlooD study demonstrated the platform's ability to reproducibly profile intra-individual lipidome phenotypes across different blood matrices and time points. Distinct lipidome patterns were observed for plasma/serum and DBS samples. A substantial number of reproducible unknown features were identified, highlighting the potential for future structural elucidation and biomarker discovery.

This study successfully developed a high-throughput 4D TIMS lipidomics platform demonstrating superior performance compared to existing methods in terms of throughput, accuracy and reproducibility. The use of stringent 4D annotation criteria substantially reduced false-positive identifications, showcasing the power of integrating multiple dimensions of data for confident lipid annotation. The consistent results across different blood matrices and time points demonstrate the robustness of the platform for clinical applications. The identification of a large number of unknown features points towards future opportunities for further biomarker discovery. While limitations exist in resolving some isomers, particularly with shorter chromatographic gradients and broader ion mobility resolution, the platform's overall performance highlights its value for large-scale clinical studies. The successful quantification of CERT2 scores demonstrates the platform's utility for clinically relevant marker profiling.

This study establishes a robust high-throughput 4D TIMS lipidomics platform for comprehensive and reproducible clinical lipid profiling. The platform's accuracy, precision, and throughput make it suitable for large-scale studies. Future work should focus on expanding the 4D library, developing improved data processing tools for resolving isomers, and conducting larger clinical studies to validate the platform's utility in various disease contexts. Standardization efforts across different lipidomic platforms are also essential for enhanced data comparability and clinical translation.

The study's limitations include the relatively small sample size in the LBlooD study and the challenges in resolving certain lipid isomers using the current chromatographic and ion mobility separation conditions. The choice of ISTDs can influence quantification results, necessitating further systematic evaluation to optimize ISTD selection for different lipid classes and disease contexts. Further work is needed to fully characterize the large number of unidentified features discovered in this study.