Breath pharmacometabolomics for therapeutic drug monitoring of epilepsy and type-1 diabetes

Background: Pharmacometabolomics is the study of pharmaceuticals and their effect on metabolites in the human body. By studying pharmacometabolomics, one can gain insights into the mode of action and effectiveness of pharmaceuticals, which is patient-dependent, and thereby infer information for personalized medicine purposes. This study can be done on various bodily fluids. One of the fastest ways to obtain this information is by analyzing breath. It provides a window into human metabolism in a non-invasive, fast, and cost-effective way. One novel approach is the use of secondary electrospray ionization coupled with high-resolution mass spectrometry (SESI-HRMS). Within this thesis, we will look at two applications of personalized medicine utilizing breath by SESI-HRMS. Namely, for epilepsy and type-1 diabetes.
Aim: Epilepsy and type-1 diabetes are two of the most common chronic diseases worldwide and are associated with tremendous medical costs and their monitoring entails invasive methods. Furthermore, the exact pathophysiology of these diseases is still debated. Previous work in our lab reported the therapeutic drug monitoring (TDM) for anti-seizure medication (ASM) valproic acid (VPA) and in addition identified metabolites to be involved in response to and side effects from ASMs by breath analysis. The identification however, was only on a low level of confidence. Therefore, in this thesis we aim to unambiguously confirm these reported metabolites. In addition, the method was done in a real-time fashion, meaning that patients have to directly exhale into an instrument, which for (cognitively) impaired patients is cumbersome. Secondly, we aim to convert this real-time method to a patient friendly off-line method. Lastly, we aim to study the anabolic effect of insulin on patients undergoing diabetic ketoacidosis (DKA) for pathophysiological understanding and monitoring.
Methods: For breath measurements, SESI-HRMS was utilized. For patients incapable of performing real-time measurements or for evaluation purposes, Nalophan bags were used to capture breath, after which their contents were infused into the SESI-HRMS. For the confirmation of metabolites in response to and side effects from ASMs, exhaled breath condensate (EBC) was used, which was subsequently analyzed by ultra-high-performance liquid chromatography coupled with HRMS. Hereafter, the data was analyzed using statistical and machine learning methods.
Results: We successfully unambiguously confirmed six amino acids in breath by utilizing EBC with a “level 1” confidence, which were previously reported to be involved in response to and side effects from ASMs. Second, the feasibility of transitioning an existing TDM approach by breath to an off-line approach was successfully assessed. Measurements taken over the course of 3.5 years showed good agreement in terms of accuracy but showed an overall decrease in signal intensity when utilizing a bag system. Furthermore, free VPA and total VPA correlated with a Lin’s concordance correlation coefficient (CCC) of 0.60 and 0.47 respectively. Metabolites reported to be involved in response to and side effects from ASM showed poor Lin’s CCC values, indicating that this information is largely lost by the bag system. Lastly, we were able to show clear metabolic trajectories in patients undergoing DKA after insulin onset. Breath analysis revealed that the generally accepted altered pathways were measurable by breath and were, to a large extent, in agreement with previously reported observations. Acetone and acetoacetate were identified as metabolites with the ability to monitor DKA progression, with breath acetoacetate being a novel finding. Finally, DKA patients demonstrated trajectories towards those of patients stratified as “in control” or “non-diabetic” after insulin onset.
Conclusion: This thesis provided novel insights into the capabilities of breath analysis to perform pharmacometabolomics in a real clinical setting. First, we identified amino acids reported to be involved in response to and side effects from ASM. This allows to confidently infer information from these metabolites in the future. Second, we showed the ability to measure free VPA and total VPA by offline SESI-HRMS, allowing bedside and (cognitively) impaired individuals to benefit from the technique. Lastly, we provided a framework for disease monitoring and pathophysiological understanding of DKA. This holds the potential to be used bedside in the Intensive care unit (ICU) for DKA disease progression as well as a further pathophysiological understanding of DKA. Overall, we showed the power of breath analysis by SESI-HRMS as a tool to perform pharmacometabolomics studies.