Suppress or Not to Suppress … CRAFT It: A Targeted Metabolomics Case Study Extracting Essential Biomarker Signals Directly from the Full 1H NMR Spectra of Horse Serum Samples

TL;DR

This study introduces a simple method called CRAFT to extract key inflammation biomarkers from NMR spectra of horse serum, avoiding complex processing steps.

Contribution

The novel contribution is a simplified, automated method using CRAFT for targeted metabolomics without advanced experimental or processing steps.

Findings

CRAFT enables direct extraction of inflammation biomarker signals from 1H NMR spectra of horse serum.

The method successfully distinguishes between cohorts with and without OCD using statistical analysis.

CRAFT-based analysis is applicable to various biofluids and can be used retrospectively.

Abstract

Background: There are a few very specific inflammation biomarkers in blood, namely lipoprotein NMe+ signals of protein clusters (GlycA and GlycB) and a composite resonance of phospholipids (SPC). The relative integrals of these resonances provide clear indication of the unique metabolic changes associated with disease, specifically inflammatory conditions, often related to serious diseases such as cancer or COVID-19 infection. Relatively complicated, yet very efficient experimental methods have been introduced recently (DIRE, JEDI) to suppress the rest of the spectrum, thus allowing measurement of these integrals of interest. Methods: In this study, we introduce a simple alternative processing method using CRAFT (Complete Reduction to Amplitude-Frequency Table), a time-domain (FID) analysis tool which can highlight selected subsets of the spectrum by choice for quantitative analysis. The output of this approach is a direct, spreadsheet-based representation of the required peak amplitude (integral) values, ready for comparative analysis, completely avoiding all the convectional data processing and manipulation steps. The significant advantage of this alternative method is that it only needs a simple water-suppressed 1D spectrum with no further experimental manipulation whatsoever. In addition, there are no pre/post processing steps (such as baseline and/or phase), further minimizing potential dependency on subjective decisions by the user and providing an opportunity to automate the entire process. Results: We applied this methodology to horse serum samples to follow the presence of inflammation for cohorts with or without OCD (Osteochondritis Dissecans) conditions and find diagnostic separation of the of the cohorts through statistical methods. Conclusions: The powerful and simple CRAFT-based approach is suitable to extract selected biomarker information from complex NMR spectra and can be similarly applied to any other biofluid from any source or sample, also retrospectively. There is a potential to extend such a simple analysis to other, previously identified relevant markers as well.

Figures