Research / NMR metabolomics & biomedical applications

NMR metabolomics under a clock

In a metabolomics experiment the spectrum is a crowd, and in a clinical one the answer has a deadline. Both constraints point at the same methodological problem.

The problem

A metabolomic sample is a mixture of dozens to hundreds of small molecules at wildly different concentrations. Their ¹H spectra overlap heavily — multiplets from unrelated compounds land on top of one another, and the resulting envelope cannot be assigned by inspection. The textbook fix is a two-dimensional experiment, which resolves the overlap beautifully and takes hours.

Hours is often fine. It is not fine when the sample is the culture medium an embryo is currently sitting in, and the clinical decision about which embryo to transfer is being made this morning. Nor is it fine when the study needs hundreds of samples measured consistently. Acquisition time stops being an inconvenience and becomes the binding constraint on whether the method is usable at all.

The approach

During my master's work at the NMR Research Centre, Indian Institute of Science, I developed phase-modulated selective methods that separate overlapping spin systems without paying the full price of a conventional 2D acquisition. The idea is to encode the information that distinguishes one spin system from another into a modulation applied during the experiment, so that a small number of one-dimensional acquisitions carries what would otherwise require a densely sampled indirect dimension.

The same line of work produced rapid data-acquisition schemes that improve resolution and sensitivity per unit time — the figure of merit that actually matters when the instrument time, not the sample, is scarce.

Schematic, not measured data. The methodological goal: recover the separation a 2D experiment would give you, at a fraction of the acquisition time, so that the method survives contact with a clinical workflow.

The application

With Satish K. Adiga's group in the Department of Clinical Embryology at Kasturba Medical College, Manipal, these methods were applied to the spent culture medium of human embryos. This is an attractive clinical target precisely because it is non-invasive — the medium is discarded anyway, so measuring it costs the embryo nothing, unlike a biopsy.

The work produced several results, including that the metabolomic signature of sibling embryos is not altered by ICSI in non-male-factor infertility, and that combining the NMR signature with embryological scoring under machine learning improves the prediction of which embryos implant successfully.

Non-invasive — measures the medium, not the embryo.
Fast enough to matter — acquisition compatible with a clinical timeline.
Combinable — metabolomic data improves prediction when fused with existing embryological assessment rather than replacing it.

A related review with the group covers the broader use of NMR in assisted reproduction (Open Biology, 2020).

Relevant papers: Anal. Chem. 94, 16555 (2022) · Magn. Reson. Chem. 59, 300 (2021) · PLoS ONE 17, e0273321 (2022) · Reprod. Sci. 30, 984 (2022).