Technical Report: A Comprehensive Comparison between Different Quantification Versions of Nightingale Health's ¹H-NMR...

Area of Science:

• Biochemistry
• Clinical Chemistry
• Bioinformatics

Background:

• Proton Nuclear Magnetic Resonance (¹H-NMR) metabolomics is vital for health and disease monitoring.
• Nightingale Health, a key ¹H-NMR metabolomics provider, updated its quantification strategy to meet clinical standards.
• Updates in quantification can impact the backward replicability of longitudinal studies.

Purpose of the Study:

• To compare Nightingale Health's ¹H-NMR metabolomics data quantification versions (2014/2016 vs. 2020).
• To assess the impact of quantification updates on analyte concordance and multi-analyte scores.
• To evaluate the replicability of previously established mortality associations.

Main Methods:

• Utilized BBMRI-NL consortium data (~28,000 samples, 28 cohorts).
• Compared original (2014/2016) and re-quantified (2020) Nightingale Health data from identical NMR spectra.
• Analyzed analyte concordance using Spearman's correlation (ρ) and evaluated multi-analyte scores (MetaboAge, MetaboHealth).

Main Results:

• High concordance observed for most analytes (73/222 with ρ > 0.9).
• Five analytes (acetoacetate, LDL-L, saturated fatty acids, S-HDL-C, sphingomyelins) showed consistently lower correlations (ρ < 0.7).
• MetaboHealth replicated well; MetaboAge required retraining due to discontinued analytes, but both scores retained mortality associations.

Conclusions:

• Nightingale Health's updated quantification strategy shows high concordance but necessitates careful consideration of specific analytes.
• Researchers should be cautious about mixing different platform versions to avoid data inconsistencies (e.g., units, discontinued analytes).
• While updated data recapitulates key findings, retraining of scores may be necessary for longitudinal reproducibility.