Abstract:Metabolite profiling of urine has seen much advancement in recent years, and its analysis by nuclear magnetic resonance (NMR) spectroscopy has become well established. However, the highly variable nature of human urine still requires improved protocols despite some standardization. In particular, diseases such as kidney disease can have a profound effect on the composition of urine and generate a highly diverse sample set for clinical studies. Large variations in pH and the cationic concentration of urine play an important role in creating positional noise within datasets generated from NMR. We demonstrate positional noise to be a confounding variable for multivariate statistical tools such as statistical total correlation spectroscopy (STOCSY), thereby hindering the process of biomarker discovery. We present a two-dimensional buffering system using potassium fluoride (KF) and phosphate buffer to reduce positional noise in metabolomic data generated from urine samples with various levels of proteinuria. KF reduces positional noise in citrate peaks, by decreasing the mean relative standard deviation (RSD) from 0.17 to 0.09. By reducing positional noise with KF, STOCSY analysis of citrate peaks saw significant improvement. We further aligned spectral data using a recursive segment-wise peak alignment (RSPA) method, which leads to further improvement of the positional noise (RSD = 0.06). These results were validated using diverse selection of metabolites which lead to an overall improvement in positional noise using the suggested protocol. In summary, we provide an improved workflow for urine metabolite biomarker discovery to achieve higher data quality for better pathophysiological understanding of human diseases.Graphical abstractCitrate peaks in the range 2.75–2.5 ppm from datasets with different sample preparation protocols and with/without in silico alignment. A Citrate peaks with standard phosphate buffering and without in silico alignment. B citrate peaks with standard phosphate buffering and with in silico alignment. C citrate peak with additional potassium fluoride and standard phosphate buffering without in silico alignment. D citrate peaks with additional potassium fluoride and standard phosphate buffering with in silico alignment. Below the respective spectrum are displayed the percent relative standard deviation (RSD) of the respective citrate peaks. This is a measure of the positional noise of peaks within a 1H NMR analysis. It can be seen that D performs the best in reducing positional noise of citrate peaks. E–H STOCSY analysis of correlating spectral features with the driver peak at 2.675 ppm (see red arrow) to identify structural correlations. As a, b, c, and d are known to be structurally correlated, STOCSY analysis should reveal r2 = 1 if data is perfectly aligned and can therefore be used as a measure of peak alignment. E Strong positional noise does not allow identifying the c and d peaks of the AB system to be correlated. F, G Neither in silico alignment or KF addition alone can completely improve the alignment and therefore increase the correlations. H Highly improved alignment by combining both KF addition and in silico alignment reduces positional noise and elucidates all four citrate peaks to be strongly correlated

Metabolomic analysis of urine samples by UHPLC-QTOF-MS: Impact of normalization strategies

Towards Standardization of Data Normalization Strategies to Improve Urinary Metabolomics Studies by GC×GC-TOFMS

Evaluation of statistical techniques to normalize mass spectrometry-based urinary metabolomics data

High-Precision Automated Workflow for Urinary Untargeted Metabolomic Epidemiology

Characteristics of Normalization Methods in Quantitative Urinary Metabolomics—Implications for Epidemiological Applications and Interpretations

Combination of Injection Volume Calibration by Creatinine and Ms Signals' Normalization to Overcome Urine Variability in Lc-Ms-Based Metabolomics Studies

Proof-of-Concept Workflow for Establishing Reference Intervals of Human Urine Proteome for Monitoring Physiological and Pathological Changes

Normalization techniques for PARAFAC modeling of urine metabolomic data

Pre-Analytical Factors that Affect Metabolite Stability in Human Urine, Plasma, and Serum: A Review

NMR Precision Metabolomics: Dynamic Peak Sum Thresholding and Navigators for Highly Standardized and Reproducible Metabolite Profiling of Clinical Urine Samples

The impact of methodological choices when developing predictive models using urinary metabolite data

MetaboGroup S: A Group Entropy-Based Web Platform for Evaluating Normalization Methods in Blood Metabolomics Data from Maintenance Hemodialysis Patients.

Influences of Normalization Method on Biomarker Discovery in Gas Chromatography-Mass Spectrometry-Based Untargeted Metabolomics: What Should Be Considered?

Metabolomic Profiling of Human Urine as a Screen for Multiple Inborn Errors of Metabolism.

Optimising a urinary extraction method for non-targeted GC–MS metabolomics

Development and Validation of an Improved Probabilistic Quotient Normalization Method for LC/MS- and NMR-based Metabonomic Analysis

A Novel Bioinformatics Approach to Identify the Consistently Well-Performing Normalization Strategy for Current Metabolomic Studies

1H NMR-based metabolite profiling workflow to reduce inter-sample chemical shift variations in urine samples for improved biomarker discovery

Evaluation of Normalization Methods for Analysis of LC-MS Data

Optimal Normalization Method for GC-MS/MS-Based Large-Scale Targeted Metabolomics

Systematic, Modifying Group-Assisted Strategy Expanding Coverage of Metabolite Annotation in Liquid Chromatography–Mass Spectrometry-Based Nontargeted Metabolomics Studies