Effects of Sample Dilution on Nuclear Magnetic Resonance-Derived Metabolic Profiles of Human Urine

Traditionally, a relatively big urine volume (e.g., 500 mu L) is used in nuclear magnetic resonance (NMR)-based human metabolomics, which is not feasible for studies with limited/ precious samples. Although urine may be diluted before conventional high-throughput metabolomics analysis, the comprehensive effect of urine dilution on metabolic profiles is unknown. Here, for the first time, we systematically investigated the effect of urine dilution on 1H NMR metabolic profiles, by evaluating signal detectability, integration, signal-to-noise ratio (SNR), chemical shift (d) and its variation, and signal overlapping of 47 metabolites in 10 volunteers. We observed significant linear changes along with increased dilution, including decreased integration and SNR, altered d, decreased intersample variation of d, and increased separation between overlapped signals, e.g., lactate and threonine, ss-D-glucose and an unassigned signal, and histidine and 3methylhistidine. We further tested the 40% dilution level (i.e., employing 300 mu L urine) in an epidemiological study containing 1018 pregnant women from the Tongji-Shuangliu Birth Cohort, showing acceptable detectability and chemical shift variability for most of the 47 metabolites profiled. It indicated that mild (e.g., 40%) dilution of human urine can largely preserve the high-abundance metabolites profiled, reduce intersample chemical shift variations, and increase separations of overlapped signals, which is an improvement of routine sample preparation methods in NMR-based metabolomics and is applicable for studies with limited urine volumes, including large-scale epidemiological studies.

Automated summary

This study systematically investigated the effect of urine dilution on 1H NMR metabolic profiles. It found that mild dilution (e.g. 40%) of human urine can preserve high-abundance metabolites, reduce intersample chemical shift variations, and increase separations of overlapped signals. This improvement is applicable for NMR-based metabolomics studies with limited urine volumes, including large-scale epidemiological studies.