Year: 2025
Type: dissertation
Study design: Observational Study
Abstract: Inborn errors of metabolism (IEM) are a diverse group of rare genetic disorders in which the body's ability to process and utilize nutrients is disrupted due to defects in specific proteins involved in metabolism. Patients present with a wide range of symptoms, from mild to life-threatening, that can impact multiple organs and systems. Early diagnosis and targeted management are essential to prevent complications and improve outcomes for those affected by these metabolic disorders. Diagnosis of inborn errors of metabolism is, however, challenging. First, due to the heterogeneity of the jointly abundant, yet individually rare diseases and phenotypes. Basic laboratory tests and clinical history, and presentation often permit targeted investigations. These consist of a wide range of methods, many of which are reliant on two techniques: mass spectrometry and spectrophotometry. One should be wary of the correct collection and handling of samples, as they are prone to environmental factors and affected by incorrect handling (e.g. storage). Functional assays, often on fibroblasts, can give additional insights into disease phenotypes. Finally, diagnosis is often confirmed by genomic testing. The diagnostic process consists of three phases: the pre-analytical phase, the analytical phase, and the post-analytical phase which includes interpretation and reporting. In the first three chapters we aimed to optimize one or more of these phases of the laboratory process. In chapter two, we described two galactosemia cases in which galactose accumulation resulted in falsely elevated levels of glucose on point-of-care glucometers using a glucose dehydrogenase technique. Central chemistry analyzers, which use the hexokinase method, are not prone to this interference, resulting in a glucose gap. Knowledge of this "glucose gap" can prompt clinicians and laboratory specialists to search for galactosemia and prevent iatrogenic damage. This consideration is an example of how post-analytical considerations can alter the clinical outcome. In chapter three, we provided an extensive overview of clinical, biochemical, and genetic findings in 93 patients with a mitochondrial HMGCS2 deficiency. HMGCS2 deficiency is a ketogenesis defect, which can result in a life-threatening episode of energy deficiency. The disease most often manifests below the age of three, during a period of decreased intake, such as a gastroenteritis. While disease-specific biomarkers have been proposed a decade ago, these metabolites are often only searched for after diagnosis, highlighting the need for easily accessible biomarkers. As such, we proposed two profiles that can be detected on routinely used methods: urine organic acid analysis and acylcarnitine analysis. In the first, a combination of proteolytic markers such as elevated isovaleric acid above the 95th percentile with dicarboxyluria in the absence of adequate ketones might be pathognomonic. In acylcarnitine analysis C2/C0 ratio is highly increased during the metabolic crisis. Additionally, we have highlighted the importance of pre-analytical factors such as the sampling time and matrix. Samples should be collected during the metabolic crises. Plasma acyl carnitine analysis is preferred over dry blood spot analysis. In chapter four, we explored the use of stable isotope tracer metabolomics, proteomics, and machine learning to analyze changes in patient samples, to develop reliable, high-throughput, and minimally invasive methods to improve diagnosis for mitochondrial diseases. Mitochondrial diseases are a genetically, biochemically, and clinically heterogeneous group, in which, despite recent advances in technologies, diagnosis remains difficult. It is estimated that more than one in three patients does not receive a definite genetic diagnosis. The use of -omics technologies as analytical tools in inborn errors of metabolism can help in interpreting variants of unknown significance or target genetic research. We included a total of 17 patients with mitochondrial diseases affecting 11 different genes, and 44 presumed healthy individuals were eligible for inclusion between March 2024 and May 2025. Patients with mitochondrial mutations were eligible for inclusion if heteroplasmy levels in leukocytes were known and exceeded 50%. Protein pattern analysis of single protein expression and respiratory complexes resulted in maximal accuracy levels of 90%. Machine learning models predicting disease status based on stable isotope metabolomics or proteomics showed high to moderate accuracy, ranging from 86% to 93%, depending on the model and technique. Our study demonstrates that the use of proteomics and/or metabolomics on a single blood sample is a highly promising diagnostic tool for mitochondrial diseases, providing valuable insights into the pathogenicity of genetic variants affecting mitochondrial function across multiple levels of the central dogma. In chapter five, we aimed to use the same technological tools to characterize the metabolic rewiring in sorbitol dehydrogenase deficiency (SORDD). This disease, discovered in 2020, is an autosomal recessive neuropathy with a prevalence of 1 in 100,000. It is characterized by slowly progressive muscle weakness, abnormal gait, and distal deformities, with or without sensory involvement. Biochemically, SORDD results in the accumulation of sorbitol and xylitol due to impaired conversion to fructose and D-xylulose. The mechanism by which SORDD causes polyneuropathy remains poorly understood. We used skin fibroblast cell cultures to investigate the (stable isotope) metabolic and proteomic alterations in SORD deficiency, aiming to better understand the impact of sorbitol and xylitol accumulation under stimulatory conditions. SORDD patient fibroblasts exhibited a phenotype marked by impaired glycolytic activity, as evidenced by decreased levels of glycolytic intermediates and a reduced contribution of glycolysis to citric acid cycle turnover. Additionally, we observed intracellular accumulation of polyol-phosphates and anhydroglucitol, along with downregulated levels of N-acetyl-aspartic acid and glycerol-3-phosphate. We hypothesize that polyol phosphates, anhydroglucitol, or phosphate trapping may play a (regulatory) role in the pathophysiology of SORDD. Further research is needed to confirm this hypothesis and validate the potential pathogenic mechanism in animal or co-culture models. Despite substantial progress in genomics and metabolomics, research into inborn errors of metabolism remains constrained by small patient cohorts, which limit statistical power and model generalizability. This thesis underscores the challenges of rare disease research, including vulnerability to individual outliers, random biological variation, and the risk of overfitting in machine learning models. Studies on HMGCS2 deficiency, SORD deficiency, and mitochondrial diseases highlight how biological heterogeneity and small sample sizes can obscure true disease signals. To address this, we implemented methodological refinements such as stricter sample standardization, selective patient inclusion, and cross-institutional collaboration. Looking forward, we position proteomics as a promising next-generation diagnostic tool for IEMs, offering functional insights that bridge the gap left by genomic uncertainty, especially in cases involving variants of unknown significance. The thesis discussed how recent technological innovations now permit high-throughput, sensitive protein profiling, even in low-input clinical samples, and how machine learning can aid in interpreting complex datasets. However, challenges such as sample variability and tissue-specific differences remain. Integrating proteomics with other -omics data, AI-powered analysis, and improved tissue models (e.g., urine-based diagnostics) could revolutionize rare disease diagnostics, enhance biomarker discovery, and deepen understanding of metabolic dysfunction in both genetic and acquired disorders.
Source: Lirias
Author: Decru, Bram
Language: English
Subfield: Clinical Biochemistry
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7111492562
Cites:
Cited by:
Related to: