Modern lipidomics can reveal molecular changes linked to disease, metabolism and treatment response, but the amount of blood needed for conventional workflows can make repeated or minimally invasive sampling difficult. A new peer-reviewed study shows that a nanoflow mass-spectrometry workflow can recover much of the detectable lipid information from just one microlitre of plasma.
Researchers evaluated a nanoLC-TIMS-TOF workflow using a standardised human plasma reference material and found that 1 µL plasma extractions retained more than 80% of the lipid features detected from 50 µL samples while using 98% less starting material. The study, published in Scientific Reports on 4 October 2026, also identified practical trade-offs involving extraction chemistry, buffer concentration, instrument pressure and sample stability.
Why smaller samples matter
Lipidomics measures large numbers of fats and fat-related molecules in biological samples. These molecules can provide information about cell membranes, energy storage, signalling and metabolic state. The field has applications in biomarker discovery and precision medicine, but analytical workflows often require sample volumes that are inconvenient when blood is scarce or needs to be collected repeatedly.
Microsampling could be particularly useful in paediatric research, longitudinal monitoring and studies where only tiny specimens are available. The analytical challenge is that reducing sample volume also reduces the amount of material reaching the instrument. At very low volumes, surface losses, pipetting variation and matrix effects can become increasingly important.
Klidel Fae Rellin and colleagues therefore tested whether nanoflow liquid chromatography combined with trapped ion mobility spectrometry and time-of-flight mass spectrometry could preserve useful lipid information when plasma input was reduced to the microlitre scale.
The researchers systematically reduced plasma input
The team used NIST SRM 1950 plasma, a widely used reference material that provides a consistent matrix for analytical comparisons. They evaluated two single-phase lipid extraction approaches, one based on methanol and methyl tert-butyl ether and another based on isopropanol and water.
Plasma input volumes ranged from 1 to 50 µL. The design included three extraction replicates and three technical injection replicates for each condition, allowing the researchers to assess the consistency of feature detection and chromatographic behaviour. Across the extraction comparison, 36 data files including blanks were analysed.
The analytical platform combined nanoflow liquid chromatography with trapped ion mobility spectrometry and time-of-flight mass spectrometry. A 1 µL aliquot of each prepared extract was injected into the nanoLC system. The method used positive-ion detection and data-dependent acquisition with parallel accumulation serial fragmentation, while ion mobility added an additional separation dimension that helped distinguish structurally similar signals.
The researchers also tested mobile phases containing 2, 5 or 10 mM ammonium formate. This mattered because salt concentration can influence chromatographic behaviour, ionisation and instrument pressure, especially in the narrow columns used for nanoflow analysis.
One microlitre preserved most detected features
The largest 50 µL samples produced the broadest feature coverage, confirming that input volume still affects profiling depth. However, reducing the starting material fiftyfold did not eliminate most of the detectable lipid information. The 1 µL extractions recovered more than 80% of the lipid features detected at 50 µL while requiring 98% less plasma.
The dominant putatively annotated lipid classes also remained represented at low volumes. Lysophosphocholines, phosphocholines and sphingomyelins were among the most frequently observed classes across both extraction protocols and the tested low-volume conditions.
That does not mean the 1 µL and 50 µL samples were analytically equivalent. Multivariate analyses showed volume-dependent differences, and the 50 µL samples separated most strongly from the lower-volume conditions. Relative class-level signal distributions also changed. For example, the 50 µL condition showed a greater relative contribution from phosphoinositol-associated features and a lower contribution from triacylglyceride and diacylglyceride-associated features than several lower-volume conditions.
The authors therefore distinguish preservation of broad class representation and feature overlap from preservation of the complete lipid profile. The study supports feasibility at very low volume, not equivalence between volumes.
Five millimolar buffer balanced feature detection and pressure stability
Buffer concentration produced another practical result. The 5 mM ammonium formate condition yielded 3,606 detected features, compared with 2,909 under the 10 mM condition. Reducing the concentration further to 2 mM did not improve feature detection under the tested conditions.
The higher salt concentration also came with a substantial pressure penalty. Over 12 hours of continuous acquisition, pressure increased by 56 bar in pump A and 52 bar in pump B under the 10 mM condition. Under the 2 and 5 mM conditions, pressure drift was no greater than 13 bar.
These results led the researchers to select 5 mM ammonium formate as a practical operating condition for their setup. Importantly, they do not claim that 5 mM is universally optimal. The comparisons were descriptive, and the experiment was not designed to establish a general buffer optimum across instruments or lipidomics applications.
Extraction chemistry affected stability during long runs
The two extraction protocols produced broadly similar class-level summaries, but they behaved differently when extracts remained in the autosampler. The researchers compared chromatographic profiles at zero, 14 and 30 hours.
Methanol and MTBE extracts showed a marked loss of signal by 14 hours and were nearly depleted of detectable lipid features after 30 hours. Isopropanol and water extracts retained greater chromatographic complexity over the same period, although signal abundance still declined.
This creates a practical trade-off. The methanol and MTBE workflow is faster to process, while the isopropanol and water method requires a two-hour incubation but appeared more suitable when samples may wait longer before analysis. For large cohorts or extended analytical sequences, stability may therefore matter as much as extraction speed.
What the findings could enable
Using only 1 µL of plasma could expand the situations in which lipidomic profiling is practical. Smaller samples may reduce the burden of repeated blood collection, make longitudinal sampling easier and allow researchers to work with populations or archived materials where specimen volume is limited.
The ion-mobility component also helped reduce spectral congestion by separating ions according to their mobility as well as mass-to-charge ratio. This provides useful additional evidence when assigning putative lipid identities, particularly for compounds with similar masses.
Still, the study is primarily an analytical feasibility investigation. It does not show that a 1 µL test can diagnose disease or replace established clinical assays. Instead, it demonstrates that substantial positive-ion lipid information can survive aggressive sample miniaturisation under a defined laboratory workflow.
Important limitations remain
The researchers evaluated positive-ion mode only, so the findings cannot automatically be extended to lipids that are better characterised in negative-ion mode. They also did not perform a complete quantitative validation covering recovery, linearity, limits of detection, limits of quantification and precision for individual lipid species.
Many reported identities were putative annotations supported by mass spectrometry and, where available, ion-mobility or fragmentation information. Counts of annotated features should therefore not be interpreted as confirmed counts of unique molecular lipids.
The experiments used a standardised reference plasma rather than a diverse clinical cohort. Real patient samples can vary substantially in lipid composition, medications, disease states and pre-analytical handling. Further validation will be needed to establish how robustly the workflow performs across those conditions.
Even with those qualifications, the study provides a practical demonstration of how analytical sensitivity can reduce the biological material required for lipidomics. Recovering more than 80% of the high-volume feature set from 98% less plasma suggests that microsample lipidomics can retain substantial molecular information, provided that extraction, buffer chemistry and run stability are carefully controlled.
Source Information
Study: Optimizing nanoLC-TIMS-TOF and lipidomics sample preparation for low volume of plasma
Authors: Klidel Fae Rellin, Lisa Abel, Aiko Barsch and colleagues
Journal: Scientific Reports
Published: 4 October 2026
DOI: 10.1038/s41598-026-72757-8
