Low-Field, Benchtop NMR Spectroscopy as a Potential Tool for Point-of-Care Diagnostics of Metabolic Conditions: Validation, Protocols and Computational Models

Novel sensing technologies for liquid biopsies offer promising prospects for the early detection of metabolic conditions through omics techniques. Indeed, high-field nuclear magnetic resonance (NMR) facilities are routinely used for metabolomics investigations on a range of biofluids in order to rapidly recognise unusual metabolic patterns in patients suffering from a range of diseases. However, these techniques are restricted by the prohibitively large size and cost of such facilities, suggesting a possible role for smaller, low-field NMR instruments in biofluid analysis. Herein we describe selected biomolecule validation on a low-field benchtop NMR spectrometer (60 MHz), and present an associated protocol for the analysis of biofluids on compact NMR instruments. We successfully detect common markers of diabetic control at low-to-medium concentrations through optimised experiments, including α-glucose (≤2.8 mmol/L) and acetone (25 µmol/L), and additionally in readily accessible biofluids, particularly human urine. We present a combined protocol for the analysis of these biofluids with low-field NMR spectrometers for metabolomics applications, and offer a perspective on the future of this technique appealing to ‘point-of-care’ applications.

SECTION S1. LF 60 MHz benchtop 1 H NMR spectra of 5.00, 8.00 and 10.00 mmol./L total glucose calibration standards. Figure S1. LF 60 MHz benchtop 1 H NMR spectra of (a) 10.00 and 8.00 mmol./L total glucose (red and blue respectively); and (b) 5.00 mmol./L total glucose. Typical spectra are shown. Abbreviations: α-Glc-C1-H, alpha-Glucose-C1-H; α-& β-Glc-C2H-C6H2, α-and β-Glucose bulk carbohydrate ring-C2H to -C6H2 protons. Typical spectra are shown. Abbreviations: Ace, acetone-CH3's; TSP, TSP-Si(CH3)3. The STN value for the acetone-CH3 functions' singlet resonance was found to be 8.8. Spectra were acquired by the methods outlined in section 3.1. For this sample we performed 512 scans (no dummy scans) with acquisition and repetition times of 6.4 and 15 s respectively, and a pulse angle of 90 o . A typical spectrum is shown. incubator at 37°C for a further 10 min. The cuvettes were then shaken to ensure homogeneity and then incubated for a further 10 min. Solutions were analysed spectrophotometrically at a wavelength of 510 nm (Evolution 60S, Thermo Scientific, UK).
For the determination of glucose concentrations from urine samples, the same method was employed with 5 µL of urine mixed with a 500 mL volume the of GOD-PAP reagent.
A linear calibration curve for glucose standards 0.50 mg/mL (2.77 mmol./L) to 2.5 mg/mL (13.8 mmol./L) was obtained, with a R 2 value of 0.990 which was employed for determining glucose concentrations in all urine samples available for testing. glucose oxidase-peroxide/4-aminophenzone/phenol method; (c) Dipstick visual colour testing system. Figure S6. Cumulative error rates expressed as a function of number of trees incorporated within the Random Forest (RF) classification model applied. The red line represents the overall error rate, whereas the blue and green lines show those for the type 2 diabetic and healthy control classifications.

Section S6. Random Forest-Metabolomics Classification of the 1 H NMR Profiles of Type 2 Diabetic and Healthy Control Participants
The RF technique was employed for classification and variable selection using the randomForest R package, with 500 trees and 7 predictor variables selected at each node subsequent to tuning. Datasets were randomly split into training and test sets containing approximately two thirds and one third of them respectively. The training set was used to build the RFs model and obtain an out-of-the-bag (OOB) error value in order to assess the performance of the classification.
Section S7: LF 60 MHz single-pulse 1 H NMR spectral profile of human blood plasma Figure S7. Partial (2.90-4.10 ppm region of) the 1 H NMR profile of human blood plasma acquired on a LF 60 MHz benchtop spectrometer. Assignments: 4, 5 and 6 correspond to the C2-H to C6-H2 bulk carbohydrate ring protons of α-and β-glucose, according to their specification in the Figure 8 legend; a range of further, less intense 1 H NMR resonances arising from other biomolecules are also readily observable in this 'carbohydrate region' of HF 1 H NMR spectra (i.e. ≥ 400 MHz) acquired on such samples, but only the most intense (glucose) signals are predominantly visible at 60 MHz operating frequency. Samples were prepared for analysis by the treatment of 600 µL of heparinised blood plasma with 65 µL of 2 H2O containing 0.05% (w/v) TSP, and 50 µL of a 1.00 mol./L phosphate buffer solution (pH 7.00) containing 0.40% (w/v) sodium azide. 1 H NMR spectra were acquired with acquisition and repeat times of 4.6 and 15 s respectively, a pulse angle of 90 o , and presaturation of the intense H2O/HOD resonance. A total of 2,048 scans were acquired. A typical spectrum is shown.