Screening for inborn errors of metabolism using untargeted metabolomics and out-of-batch controls

Authors:
  • MBMichiel Bongaerts
    Michiel Bongaerts
  • RBRamon Bonte
    Ramon Bonte
  • SDSerwet Demirdas
    Serwet Demirdas
  • EJEd H. Jacobs
    Ed H. Jacobs
  • EOE. Oussoren
    E. Oussoren
  • APAns T. van der Ploeg
    Ans T. van der Ploeg
  • MWMargreet A.E.M. Wagenmakers
    Margreet A.E.M. Wagenmakers
  • RHRobert M.W. Hofstra
    Robert M.W. Hofstra
  • HBHenk J. Blom
    Henk J. Blom
  • MRMarcel J.T. Reinders
    Marcel J.T. Reinders
  • GRGeorge J. G. Ruijter
    George J. G. Ruijter

Abstract
Motivation: Untargeted metabolomics is an emerging technology in the laboratory diagnosis of inborn errors of metabolism (IEM). In order to judge if metabolite levels are abnormal, analysis of a large number of reference samples is crucial to correct for variations in metabolite concentrations resulting from factors such as diet, age and gender. However, a large number of controls requires the use of out-of-batch controls, which is hampered by the semi-quantitative nature of untargeted metabolomics data, i.e. technical variations between batches. Methods to merge and accurately normalize data from multiple batches are urgently needed. Methods & results: Based on six metrics, we compared existing normalization methods on their ability to reduce batch effects from eight independently processed batches. Many of those showed marginal performances, which motivated us to develop Metchalizer , a normalization method which uses 17 stable isotope-labeled internal standards and a mixed effect model. In addition, we propose a regression model with age- and sex as covariates fitted on control samples obtained from all eight batches. Metchalizer applied on log-transformed data showed the most promising performance on batch effect removal as well as in the detection of 178 known biomarkers across 45 IEM patient samples and performed at least similar to an approach using 15 within-batch controls. Furthermore, our regression model indicates that 10-24% of the considered features showed significant age-dependent variations. Conclusions: Our comprehensive comparison of normalization methods showed that our Log-Metchalizer approach enables the use out-of-batch controls to establish clinically-relevant reference values for metabolite concentrations. These findings opens possibilities to use large scale out-of-batch control samples in a clinical setting, increasing throughput and detection accuracy. Availability: Metchalizer is available at https://github.com/mbongaerts/Metchalizer/
All Related Versions

This publication has 19 references indexed in Scilit: