Perspective: use and reuse of NMR-based metabolomics data: what works and what remains challenging

BACKGROUND

The National Cancer Institute issued a Request for Information (RFI; NOT-CA-23-007) in October 2022, soliciting input on using and reusing metabolomics data. This RFI aimed to gather input on best practices for metabolomics data storage, management, and use/reuse.

AIM OF REVIEW

The nuclear magnetic resonance (NMR) Interest Group within the Metabolomics Association of North America (MANA) prepared a set of recommendations regarding the deposition, archiving, use, and reuse of NMR-based and, to a lesser extent, mass spectrometry (MS)-based metabolomics datasets. These recommendations were built on the collective experiences of metabolomics researchers within MANA who are generating, handling, and analyzing diverse metabolomics datasets spanning experimental (sample handling and preparation, NMR/MS metabolomics data acquisition, processing, and spectral analyses) to computational (automation of spectral processing, univariate and multivariate statistical analysis, metabolite prediction and identification, multi-omics data integration, etc.) studies.

KEY SCIENTIFIC CONCEPTS OF REVIEW

We provide a synopsis of our collective view regarding the use and reuse of metabolomics data and articulate several recommendations regarding best practices, which are aimed at encouraging researchers to strengthen efforts toward maximizing the utility of metabolomics data, multi-omics data integration, and enhancing the overall scientific impact of metabolomics studies.

Mapping of 1 H NMR chemical shifts relationship with chemical similarities for the acceleration of metabolic profiling: Application on blood products

One-dimensional (1D) proton-nuclear magnetic resonance (1 H-NMR) spectroscopy is an established technique for the deconvolution of complex biological sample types via the identification/quantification of small molecules. It is highly reproducible and could be easily automated for small to large-scale bioanalytical, epidemiological, and in general metabolomics studies. However, chemical shift variability is a serious issue that must still be solved in order to fully automate metabolite identification. Herein, we demonstrate a strategy to increase the confidence in assignments and effectively predict the chemical shifts of various NMR signals based upon the simplest form of statistical models (i.e., linear regression). To build these models, we were guided by chemical homology in serum/plasma metabolites classes (i.e., amino acids and carboxylic acids) and similarity between chemical groups such as methyl protons. Our models, built on 940 serum samples and validated in an independent cohort of 1,052 plasma-EDTA spectra, were able to successfully predict the 1 H NMR chemical shifts of 15 metabolites within ~1.5 linewidths (Δv1/2 ) error range on average. This pilot study demonstrates the potential of developing an algorithm for the accurate assignment of 1 H NMR chemical shifts based solely on chemically defined constraints.

Comparison of phenomics and cfDNA in a large breast screening population: the Breast Screening and Monitoring Study (BSMS)

To assess their roles in breast cancer diagnostics, we aimed to compare plasma cell-free DNA (cfDNA) levels with the circulating metabolome in a large breast screening cohort of women recalled for mammography, including healthy women and women with mammographically detected breast diseases, ductal carcinoma in situ and invasive breast cancer: the Breast Screening and Monitoring Study (BSMS). In 999 women, plasma was analyzed by nuclear magnetic resonance (NMR) and Ultra-Performance Liquid Chromatography-Mass Spectrometry (UPLC-MS) and then processed to isolate and quantify total cfDNA. NMR and UPLC-MS results were compared with data for 186 healthy women derived from the AIRWAVE cohort. Results showed no significant differences between groups for all metabolites, whereas invasive cancers had significantly higher plasma cfDNA levels than all other groups. When stratified the supervised OPLS-DA analysis and total cfDNA concentration showed high discrimination accuracy between invasive cancers and the disease/medication-free subjects. Furthermore, comparison of OPLS-DA data for invasive breast cancers with the AIRWAVE cohort showed similar discrimination between breast cancers and healthy controls. This is the first report of agreement between metabolomics and plasma cfDNA levels for discriminating breast cancer from healthy subjects in a true screening population. It also emphasizes the importance of sample standardization. Follow on studies will involve analysis of candidate features in a larger validation series as well as comparing results with serial plasma samples taken at the next routine screening mammography appointment. The findings here help establish the role of plasma analysis in the diagnosis of breast cancer in a large real-world cohort.

Optimization of Diffusion-Ordered NMR Spectroscopy Experiments for High-Throughput Automation in Human Metabolic Phenotyping

The diffusion-ordered nuclear magnetic resonance spectroscopy (DOSY) experiment allows the calculation of diffusion coefficient values of metabolites in complex mixtures. However, this experiment has not yet been broadly used for metabolic profiling due to lack of a standardized protocol. Here we propose a pipeline for the DOSY experimental setup and data processing in metabolic phenotyping studies. Due to the complexity of biological samples, three experiments (a standard DOSY, a relaxation-edited DOSY, and a diffusion-edited DOSY) have been optimized to provide DOSY metabolic profiles with peak-picked diffusion coefficients for over 90% of signals visible in the one-dimensional 1H general biofluid profile in as little as 3 min 36 s. The developed parameter sets and tools are straightforward to implement and can facilitate the use of DOSY for metabolic profiling of human blood plasma and urine samples.

NMRpQuant: an automated software for large scale urinary total protein quantification by one-dimensional 1H NMR profiles

SUMMARY

1H nuclear magnetic resonance (NMR) spectroscopy is an established bioanalytical technology for metabolic profiling of biofluids in both clinical and large-scale population screening applications. Recently, urinary protein quantification has been demonstrated using the same 1D 1H NMR experimental data captured for metabolic profiling. Here, we introduce NMRpQuant, a freely available platform that builds on these findings with both novel and further optimized computational NMR approaches for rigorous, automated protein urine quantification. The results are validated by interlaboratory comparisons, demonstrating agreement with clinical/biochemical methodologies, pointing at a ready-to-use tool for routine protein urinalyses.

AVAILABILITY AND IMPLEMENTATION

NMRpQuant was developed on MATLAB programming environment. Source code and Windows/macOS compiled applications are available at https://github.com/pantakis/NMRpQuant, and working examples are available at https://doi.org/10.6084/m9.figshare.18737189.v1.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

NMR-Based Metabolomics in Differential Diagnosis of Chronic Kidney Disease (CKD) Subtypes

Chronic Kidney Disease (CKD) is considered as a major public health problem as it can lead to end-stage kidney failure, which requires replacement therapy. A prompt and accurate diagnosis, along with the appropriate treatment, can delay CKD’s progression, significantly. Herein, we sought to determine whether CKD etiology can be reflected in urine metabolomics during its early stage. This is achieved through the analysis of the urine metabolic fingerprint from 108 CKD patients by means of Nuclear Magnetic Resonance (NMR) spectroscopy metabolomic analysis. We report the first NMR—metabolomics data regarding the three most common etiologies of CKD: Chronic Glomerulonephritis (IgA and Membranous Nephropathy), Diabetic Nephropathy (DN) and Hypertensive Nephrosclerosis (HN). Analysis aided a moderate glomerulonephritis clustering, providing characterization of the metabolic fluctuations between the CKD subtypes and control disease. The urine metabolome of IgA Nephropathy reveals a specific metabolism, reflecting its different etiology or origin and is useful for determining the origin of the disease. In contrast, urine metabolomes from DN and HN patients did not reveal any indicative metabolic pattern, which is consistent with their fused clinical phenotype. These findings may contribute to improving diagnostics and prognostic approaches for CKD, as well as improving our understanding of its pathology.

1H NMR Signals from Urine Excreted Protein Are a Source of Bias in Probabilistic Quotient Normalization

Normalization to account for variation in urinary dilution is crucial for interpretation of urine metabolic profiles. Probabilistic quotient normalization (PQN) is used routinely in metabolomics but is sensitive to systematic variation shared across a large proportion of the spectral profile (>50%). Where ¹H nuclear magnetic resonance (NMR) spectroscopy is employed, the presence of urinary protein can elevate the spectral baseline and substantially impact the resulting profile. Using ¹H NMR profile measurements of spot urine samples collected from hospitalized COVID-19 patients in the ISARIC 4C study, we determined that PQN coefficients are significantly correlated with observed protein levels (r² = 0.423, p < 2.2 × 10^-16). This correlation was significantly reduced (r² = 0.163, p < 2.2 × 10^-16) when using a computational method for suppression of macromolecular signals known as small molecule enhancement spectroscopy (SMolESY) for proteinic baseline removal prior to PQN. These results highlight proteinuria as a common yet overlooked source of bias in ¹H NMR metabolic profiling studies which can be effectively mitigated using SMolESY or other macromolecular signal suppression methods before estimation of normalization coefficients.

A Computationally Lightweight Algorithm for Deriving Reliable Metabolite Panel Measurements from 1D 1H NMR

Small Molecule Enhancement SpectroscopY (SMolESY) was employed to develop a unique and fully automated computational solution for the assignment and integration of 1H nuclear magnetic resonance (NMR) signals from metabolites in challenging matrices containing macromolecules (herein blood products). Sensitive and reliable quantitation is provided by instant signal deconvolution and straightforward integration bolstered by spectral resolution enhancement and macromolecular signal suppression. The approach is highly efficient, requiring only standard one-dimensional 1H NMR spectra and avoiding the need for sample preprocessing, complex deconvolution, and spectral baseline fitting. The performance of the algorithm, developed using >4000 NMR serum and plasma spectra, was evaluated using an additional >8800 spectra, yielding an assignment accuracy greater than 99.5% for all 22 metabolites targeted. Further validation of its quantitation capabilities illustrated a reliable performance among challenging phenotypes. The simplicity and complete automation of the approach support the application of NMR-based metabolite panel measurements in clinical and population screening applications.

Isolation of Proximal Fluids to Investigate the Tumor Microenvironment of Pancreatic Adenocarcinoma

Pancreatic adenocarcinoma (PDAC) is the fourth leading cause of cancer-related death, and soon to become the second. There is an urgent need of variables associated to specific pancreatic pathologies to help preoperative differential diagnosis and patient profiling. Pancreatic juice is a relatively unexplored body fluid, which, due to its close proximity to the tumor site, reflects changes in the surrounding tissue. Here we describe in detail the intraoperative collection procedure. Unfortunately, translating pancreatic juice collection to murine models of PDAC, to perform mechanistic studies, is technically very challenging. Tumor interstitial fluid (TIF) is the extracellular fluid, outside blood and plasma, which bathes tumor and stromal cells. Similarly to pancreatic juice, for its property to collect and concentrate molecules that are found diluted in plasma, TIF can be exploited as an indicator of microenvironmental alterations and as a valuable source of disease-associated biomarkers. Since TIF is not readily accessible, various techniques have been proposed for its isolation. We describe here two simple and technically undemanding methods for its isolation: tissue centrifugation and tissue elution.

SMolESY: an efficient and quantitative alternative to on-instrument macromolecular 1H-NMR signal suppression

One-dimensional (1D) proton-nuclear magnetic resonance (1H-NMR) spectroscopy is an established technique for measuring small molecules in a wide variety of complex biological sample types. It is demonstrably reproducible, easily automatable and consequently ideal for routine and large-scale application. However, samples containing proteins, lipids, polysaccharides and other macromolecules produce broad signals which overlap and convolute those from small molecules. NMR experiment types designed to suppress macromolecular signals during acquisition may be additionally performed, however these approaches add to the overall sample analysis time and cost, especially for large cohort studies, and fail to produce reliably quantitative data. Here, we propose an alternative way of computationally eliminating macromolecular signals, employing the mathematical differentiation of standard 1H-NMR spectra, producing small molecule-enhanced spectra with preserved quantitative capability and increased resolution. Our approach, presented in its simplest form, was implemented in a cheminformatic toolbox and successfully applied to more than 3000 samples of various biological matrices rich or potentially rich with macromolecules, offering an efficient alternative to on-instrument experimentation, facilitating NMR use in routine and large-scale applications.

A community-built calibration system: The case study of quantification of metabolites in grape juice by qNMR spectroscopy

Nuclear Magnetic Resonance (NMR) is an analytical technique extensively used in almost every chemical laboratory for structural identification. This technique provides statistically equivalent signals in spite of using spectrometer with different hardware features and is successfully used for the traceability and quantification of analytes in food samples. Nevertheless, to date only a few internationally agreed guidelines have been reported on the use of NMR for quantitative analysis. The main goal of the present study is to provide a methodological pipeline to assess the reproducibility of NMR data produced for a given matrix by spectrometers from different manufacturers, with different magnetic field strengths, age and hardware configurations. The results have been analyzed through a sequence of chemometric tests to generate a community-built calibration system which was used to verify the performance of the spectrometers and the reproducibility of the predicted sample concentrations.

Metabolome of Pancreatic Juice Delineates Distinct Clinical Profiles of Pancreatic Cancer and Reveals a Link between Glucose Metabolism and PD-1+ Cells

Better understanding of pancreatic diseases, including pancreatic ductal adenocarcinoma (PDAC), is an urgent medical need, with little advances in preoperative differential diagnosis, preventing rational selection of therapeutic strategies. The clinical management of pancreatic cancer patients would benefit from the identification of variables distinctively associated with the multiplicity of pancreatic disorders. We investigated, by ¹H nuclear magnetic resonance, the metabolomic fingerprint of pancreatic juice (the biofluid that collects pancreatic products) in 40 patients with different pancreatic diseases. Metabolic variables discriminated PDAC from other less aggressive pancreatic diseases and identified metabolic clusters of patients with distinct clinical behaviors. PDAC specimens were overtly glycolytic, with significant accumulation of lactate, which was probed as a disease-specific variable in pancreatic juice from a larger cohort of 106 patients. In human PDAC sections, high expression of the glucose transporter GLUT-1 correlated with tumor grade and a higher density of PD-1⁺ T cells, suggesting their accumulation in glycolytic tumors. In a preclinical model, PD-1⁺ CD8 tumor-infiltrating lymphocytes differentially infiltrated PDAC tumors obtained from cell lines with different metabolic consumption, and tumors metabolically rewired by knocking down the phosphofructokinase (Pfkm) gene displayed a decrease in PD-1⁺ cell infiltration. Collectively, we introduced pancreatic juice as a valuable source of metabolic variables that could contribute to differential diagnosis. The correlation of metabolic markers with immune infiltration suggests that upfront evaluation of the metabolic profile of PDAC patients could foster the introduction of immunotherapeutic approaches for pancreatic cancer.

High-Throughput Metabolomics by 1D NMR

Metabolomics deals with the whole ensemble of metabolites (the metabolome). As one of the -omic sciences, it relates to biology, physiology, pathology and medicine; but metabolites are chemical entities, small organic molecules or inorganic ions. Therefore, their proper identification and quantitation in complex biological matrices requires a solid chemical ground. With respect to for example, DNA, metabolites are much more prone to oxidation or enzymatic degradation: we can reconstruct large parts of a mammoth's genome from a small specimen, but we are unable to do the same with its metabolome, which was probably largely degraded a few hours after the animal's death. Thus, we need standard operating procedures, good chemical skills in sample preparation for storage and subsequent analysis, accurate analytical procedures, a broad knowledge of chemometrics and advanced statistical tools, and a good knowledge of at least one of the two metabolomic techniques, MS or NMR. All these skills are traditionally cultivated by chemists. Here we focus on metabolomics from the chemical standpoint and restrict ourselves to NMR. From the analytical point of view, NMR has pros and cons but does provide a peculiar holistic perspective that may speak for its future adoption as a population-wide health screening technique.

NMR-based metabolomics identifies patients at high risk of death within two years after acute myocardial infarction in the AMI-Florence II cohort

BACKGROUND

Risk stratification and management of acute myocardial infarction patients continue to be challenging despite considerable efforts made in the last decades by many clinicians and researchers. The aim of this study was to investigate the metabolomic fingerprint of acute myocardial infarction using nuclear magnetic resonance spectroscopy on patient serum samples and to evaluate the possible role of metabolomics in the prognostic stratification of acute myocardial infarction patients.

METHODS

In total, 978 acute myocardial infarction patients were enrolled in this study; of these, 146 died and 832 survived during 2 years of follow-up after the acute myocardial infarction. Serum samples were analyzed via high-resolution ¹H-nuclear magnetic resonance spectroscopy and the spectra were used to characterize the metabolic fingerprint of patients. Multivariate statistics were used to create a prognostic model for the prediction of death within 2 years after the cardiovascular event.

RESULTS

In the training set, metabolomics showed significant differential clustering of the two outcomes cohorts. A prognostic risk model predicted death with 76.9% sensitivity, 79.5% specificity, and 78.2% accuracy, and an area under the receiver operating characteristics curve of 0.859. These results were reproduced in the validation set, obtaining 72.6% sensitivity, 72.6% specificity, and 72.6% accuracy. Cox models were used to compare the known prognostic factors (for example, Global Registry of Acute Coronary Events score, age, sex, Killip class) with the metabolomic random forest risk score. In the univariate analysis, many prognostic factors were statistically associated with the outcomes; among them, the random forest score calculated from the nuclear magnetic resonance data showed a statistically relevant hazard ratio of 6.45 (p = 2.16×10^-16). Moreover, in the multivariate regression only age, dyslipidemia, previous cerebrovascular disease, Killip class, and random forest score remained statistically significant, demonstrating their independence from the other variables.

CONCLUSIONS

For the first time, metabolomic profiling technologies were used to discriminate between patients with different outcomes after an acute myocardial infarction. These technologies seem to be a valid and accurate addition to standard stratification based on clinical and biohumoral parameters.

Fingerprinting Acute Digestive Diseases by Untargeted NMR Based Metabolomics

Precision medicine may significantly contribute to rapid disease diagnosis and targeted therapy, but relies on the availability of detailed, subject specific, clinical information. Proton nuclear magnetic resonance (¹H–NMR) spectroscopy of body fluids can extract individual metabolic fingerprints. Herein, we studied 64 patients admitted to the Florence main hospital emergency room with severe abdominal pain. A blood sample was drawn from each patient at admission, and the corresponding sera underwent ¹H–NMR metabolomics fingerprinting. Unsupervised Principal Component Analysis (PCA) analysis showed a significant discrimination between a group of patients with symptoms of upper abdominal pain and a second group consisting of patients with diffuse abdominal/intestinal pain. Prompted by this observation, supervised statistical analysis (Orthogonal Partial Least Squares–Discriminant Analysis (OPLS-DA)) showed a very good discrimination (>90%) between the two groups of symptoms. This is a surprising finding, given that neither of the two symptoms points directly to a specific disease among those studied here. Actually herein, upper abdominal pain may result from either symptomatic gallstones, cholecystitis, or pancreatitis, while diffuse abdominal/intestinal pain may result from either intestinal ischemia, strangulated obstruction, or mechanical obstruction. Although limited by the small number of samples from each of these six conditions, discrimination of these diseases was attempted. In the first symptom group, >70% discrimination accuracy was obtained among symptomatic gallstones, pancreatitis, and cholecystitis, while for the second symptom group >85% classification accuracy was obtained for intestinal ischemia, strangulated obstruction, and mechanical obstruction. No single metabolite stands up as a possible biomarker for any of these diseases, while the contribution of the whole ¹H–NMR serum fingerprint seems to be a promising candidate, to be confirmed on larger cohorts, as a first-line discriminator for these diseases.

Deconvoluting interrelationships between concentrations and chemical shifts in urine provides a powerful analysis tool

The NMR chemical shifts of a substance in a complex mixture strongly depend on the composition of the mixture itself, as many weak interactions occur that are hardly predictable. Chemical shift variability is the major obstacle to automatically assigning, and subsequently quantitating, metabolite signals in body fluids, particularly urine. Here we demonstrate that the chemical shifts of signals in urine are actually predictable. This is achieved by constructing ca. 4000 artificial mixtures where the concentrations of 52 most abundant urine metabolites-including 11 inorganic ions-are varied, to sparsely but efficiently populate an N-dimensional concentration matrix. A strong relationship is established between the concentration matrix and the chemical shift matrix, so that chemical shifts of > 90 metabolite signals can be accurately predicted in real urine samples. The concentrations of the invisible inorganic ions are also accurately predicted, along with those of albumin and of several other abundant urine components.

Unscrambling micro-solvation of –COOH and –NH groups in neat dimethyl sulfoxide: insights from 1H-NMR spectroscopy and computational studies

This study investigates the interactions of –COOH and –NH groups in neat DMSO solutions, with special focus on their thermodynamics and kinetics.

1H-NMR based metabolomics study for the detection of the human urine metabolic profile effects of Origanum dictamnus tea ingestion

NMR based metabolomics clarify theOriganum dictamnustea effect upon the human urine metabolome.