Benchmarking untargeted metabolomics data quality with allopurinol-induced perturbations

Abstract Introduction We present a simple test to assess whether a metabolomics dataset is fit-for-purpose. Current qualitycontrol approaches do not directly evaluate the ability to recover biologically meaningful perturbations. Objectives To evaluate whether known drug-induced metabolic perturbations can serve as internal benchmarks fordataset quality. Methods In a study (the TROMBOLOME study, unrelated to allopurinol therapy), 1,000 serum samples were analyzedwith one targeted and two untargeted metabo lomics panels. Samples were classified as allopurinol-positive (N=19)using detection of allopurinol analytical targets. Endogenous metabolite markers of allopurinol therapy wereevaluated based on hypotheses derived from the literature. Statistical evaluation was performed using Mann–Whitney U-tests. Results The hypothesis of upregulation was supported for xanthine, orotate, and orotidine (p < 0.0001) inallopurinol-positive cases (N = 19). These findings demonstrate repro ducibility of well-characterized metabolicperturbations within the dataset. Conclusion In the absence of external quality assessment schemes for untargeted metabolomics, such benchmarkscould provide a practical way to evaluate whether datasets are suitable for downstream biological interpretation.The proposed targeted exposomics approach complements traditional QC metrics by assessing biologicalrecoverability.