Fast and accurate quantitative organic acid analysis with LC-QTOF/MS facilitates screening of patients for inborn errors of metabolism

Urinary organic acid analysis is a pivotal technique in selective screening for inborn errors of metabolism (IEMs) (Tanaka et al 1980; de Almeida and Duran 2014). The current state of the art relies on gas-chromatography coupled to mass spectrometry (GC-MS) of derivatized compounds. By its nature, the literally hundreds of organic acids present in human urine, endogenous as well as microbiome, drugs and other xenobiotic-derived metabolites (Blau et al 2014), can be detected by GC-MS. In fact, GC-MS analysis is the forerunner of untargeted metabolomics analysis as we nowadays envision it.

GC-MS has several advantages. It has high separation efficiency of metabolites, high specificity and sensitivity, few matrix effects, and broadly covered mass spectra libraries for identification of metabolites of interest are widely available (Pasikanti et al 2008). Disadvantages are clearly present since organic acids are not volatile and require organic solvent extraction and derivatization prior to GC-MS analysis. This makes the procedure laborious, and compared to underivatized tandem-mass spectrometry methods, time-consuming in terms of analytical run-time. Moreover, a relatively large sample volume is needed, despite efforts to increase throughput and reduce sample volume (Nakagawa et al 2010).

High resolution proton nuclear magnetic resonance (NMR) spectroscopy is a good alternative for organic acid analysis (and even broader IEM screening) with minimal sample preparation and a short experimental time. NMR spectroscopy did, however, not evolve in common IEM screening because of financial constraints and its relatively low sensitivity in the low millimolar range (Moolenaar et al 2003). Alternative liquid chromatography (LC)-quantitative hyphenated tandem mass spectrometry (MS/MS) techniques have been developed that allow high-throughput (Want et al 2010; Bouatra et al 2013). LC-MS/MS has outstanding sensitivity (in low nanomolar range) and specificity but only targeted metabolites, i.e., those a-priori selected in the method, are detected and quantified. The relatively recent introduction of high-resolution (HR) mass spectrometry in the form of time-of flight (TOF) MS and Orbitrap MS specificity allowed a major breakthrough. LC-HR MS combines the analytical power of LC-MS/MS with the unbiased quality of classical GC-MS, and thus enables not only the quantification of target metabolites, but also facilitates untargeted metabolite screening. Until now, LC-HR MS, including LC-QTOF/MS that combined with a quadrupole (Q), has been widely deployed in research settings (Paglia et al 2012), including inborn errors of metabolism (Wikoff et al 2007). Apart from a qualitative untargeted metabolomics approach (Miller et al 2015), no quantitative application suitable for routine diagnostic setting has been reported in the inborn errors of metabolism (IEM) field.

Here, we present our newly developed LC-QTOF/MS method for the quantitative analysis of urinary organic acids. This method covers a panel of critical biomarkers targeting defects in branched-chain amino acid-, lysine- and tryptophan-, aromatic amino acid-, neurotransmitter-, fatty acid- and pyrimidine metabolism as well as disorders in the Krebs cycle and urea cycle, amino acylase deficiencies and various other disorders. In addition, four medication-related metabolites were included in the panel, to allow discrimination between metabolic defects and medication. Analytical and diagnostic suitability were demonstrated in a cohort of individual urine samples, including proven IEMs. To expedite clinical interpretation we introduced z-score value plots. Our results demonstrate the suitability of this new method in the routine setting of selective metabolic screening.

Urine organic acid analysis is a pivotal part in the diagnostic workup of inborn errors of metabolism (IEM). We developed a LC-QTOF/MS method for the quantification of 71 metabolites in urine covering all disorders classically identified through GC-MS analysis of organic acids. This targeted analyses is aimed at finding most organic acidurias in a fast manner, with the limitation that some IEM classically detected in GC-MS analysis, such as glycerol kinase deficiency, are not detected. Obviously, in case of strong clinical suspicion additional analyses are required in these cases. However, our method can be easily expanded by adding more relevant biomarkers for other IEM. The method was validated according to Dutch NEN-EN-ISO 15189 standards underlining its fit for purpose in the routine setting of a specialized diagnostic laboratory in the field of IEM. Performance characteristics (i.e., recovery) of our new analytical UHPLC-QTOF/MS assay were comparable or even superior to the classical GC-MS analysis. Cross validation showed on linearities and recoveries that our results compared well to that of the median of all labs for the GC-MS method. The targeted approach with internal standards makes our method robust. In addition, simple sample preparation and short time from unprocessed urine sample-to-authorized lab result of an individual sample (< 3–4 h) are clear benefits for the LC-QTOF/MS assay. However, for two metabolites, i.e., oxoglutaric acid and mevalonic acid, LC-QTOF/MS had a worse recovery than the classical assay. Two cases of classical mevalonic aciduria (MEVA, OMIM #610377) were easily diagnosed based on z-score profiles with abnormal mevalonic acid. Hyper-IgD syndrome (OMIM #260920) diagnostics depends on detecting more subtle increased excretion of mevalonic acid especially in periods of episodic fever (Prietsch et al 2003). We measure only mevalonic acid which is in equilibrium with lactone form (not detected in our method) in mild acid conditions; additional HIDS patient sample (especially in crisis) analysis would be required to evaluate the feasibility of these diagnostics with our new method.

We clinically validated our new assay by evaluating 32 different IEM and in total 187 samples with a representative number of control samples showed that this new diagnostic approach with z-score profiles facilitates diagnostics of the majority of organic acidurias. Most biomarkers associated with the IEMs included as reported in literature (Blau et al 2014) were observed in significantly elevated levels (z-score > 3.0) in the z-score plots for most IEM. Not all metabolites were increased to the same extent in all patients reflecting severity of the disease and clinical condition and/or treatment status (detailed information however was not always available). In patients with MSUD, not all markers were elevated in all patients; however, due to the presence of multiple biomarkers for MSUD in the metabolite panel, this diagnosis could not be missed. Glutaric acid was not elevated in all urine samples from GA I and MADD patients. With urine organic acid analysis, 3-hydroxyglutaric acid is the diagnostic metabolite. Glutaric acid can be completely normal in some patients but 3-hydroxyglutaric acid is pathognomonic for GA I. 3-hydroxyglutaric acid can be hard to identify and quantitate since it may co-elute with 2-hydroxyglutaric acid in GC-MS analyses (Hedlund et al 2006), depending on the analytical column. In our newly established LC-QTOF/MS assay, we could separate 2- and 3-hydroxyglutaric acid quite well, which allows unambiguous annotation and quantification of both metabolites.

All MCADD patients had abnormal z-score profiles. MCADD diagnostics is usually based on acylcarnitine profiling and organic acid analysis. Browning et al (2005) showed that normal acylcarnitine levels during confirmation of abnormal newborn screening have been encountered in some fatty acid oxidation disorders. Our findings are therefore of extra information on diagnostics. For these notoriously difficult IEMs several analyses (both acylcarnitine and organic acid analysis) in different settings (fasting and fed state) could be of added value in the diagnostics. Eventually, enzyme or gene analysis confirms a diagnosis based on biochemical abnormalities.

One patient with 5-oxoprolinase deficiency (pyroglutamic aciduria) could be easily discriminated on the basis of the z-score for secondary causes of 5-oxoprolinuria like a certain drug (paracetamol) (Saudubray et al 2016). The absence of paracetamol related markers in this patient supported the diagnosis. Other causes of 5-oxoprolinuria which include severe burns, inborn errors of metabolism not involving the gamma-glutamyl cycle, e.g., X-linked ornithine transcarbamylase deficiency, urea cycle defects, tyrosinemia, and homocystinuria were, however, not included in this evaluation.

All these factors and representation of results as a long list of numbers made it easy to miss a diagnosis despite reference values. This is why we introduced z-score plots. In the broad panel of IEMs included in this clinical validation, the z-score plot proved to be a useful and easy to use tool for labs to interpret the analytical results of the LC-QTOF/MS method.

The targeted biomarkers approach and number of IEM diseases that can be diagnosed can be continuously expanded to meet the complexity of IEM diagnostics. The full scan method is flexible to include new analytes and update the method to meet this challenge. This is part of our future work. We will explore not only the negative but also the positive ionization mode to cover broad chemical groups, including amino acids, acylcarnitines, purine and pyrimidine metabolites. Eventually, the method can be a generic metabolic screening method on the LC-QTOF/MS platform whereby urine is analyzed as a front-line specimen for screening of IEMs. This would provide a valuable means to efficiently, accurately, and rapidly identify and manage many IEMs (Campeau et al 2008).