Tsuyoshi Nakai

<jats:title>Abstract</jats:title><jats:p>Teicoplanin is an important antimicrobial agent for methicillin-resistant<jats:italic>Staphylococcus aureus</jats:italic>infections. To enhance its clinical effectiveness while preventing adverse effects, therapeutic drug monitoring (TDM) of teicoplanin trough concentration is recommended. Given the importance of the early achievement of therapeutic concentrations for treatment success, initial dosing regimens are deliberately designed based on patient information.</jats:p><jats:p>Considerable effort has been dedicated to developing an optimal initial dose plan for specific populations; however, comprehensive strategies for tailoring teicoplanin dosing have not been successfully implemented. The initial dose planning of teicoplanin is conducted at the clinician’s discretion and is thus strongly dependent on the clinician’s experience and expertise.</jats:p><jats:p>The present study aimed to use a machine learning (ML) approach to integrate clinicians’ knowledge into a predictive model for initial teicoplanin dose planning. We first confirmed that dose planning by pharmacists dedicated to TDM (hereafter TDM pharmacists) significantly improved early therapeutic target attainment for patients without an intensive care unit or high care unit stay, providing the first evidence that dose planning of teicoplanin by experienced clinicians enhances early teicoplanin therapeutic exposure. Next, we used a dataset of teicoplanin initial dose planning by TDM pharmacists to train and implement the model, yielding a model that emulated TDM pharmacists’ decision-making for dosing. We further applied ML to cases without TDM pharmacist dose planning and found that the target attainment rate of the initial teicoplanin concentration markedly increased. Our study opens a new avenue for tailoring the initial dosing regimens of teicoplanin using a TDM pharmacist-trained ML system.</jats:p><jats:sec><jats:title>Importance</jats:title><jats:p>Teicoplanin is used for treating methicillin-resistant<jats:italic>Staphylococcus aureus</jats:italic>infections. Given the importance of early adequate teicoplanin exposure, initial dosing regimens are adjusted for patient characteristics. However, tailoring teicoplanin dosing is challenging for most clinicians. In this study, we first showed that initial dosing regimens by pharmacists dedicated to therapeutic drug monitoring significantly improved early achievement of targeted concentration. In addition, we leveraged machine learning approach to develop the predictive model that tailors initial dosing regimens at the levels of experienced pharmacists. The target attainment rate of patients without experienced pharmacists’ dose planning was significantly increased by applying the model. Therefore, machine learning approach may provide new avenues for tailoring initial teicoplanin dosing.</jats:p></jats:sec>

ABSTRACT Objective Cerebrospinal fluid (CSF) cell‐free mitochondrial DNA (cf‐mtDNA) is a potential biomarker for Parkinson's disease (PD), but its clinical relevance remains unclear. We investigated associations between CSF cf‐mtDNA levels, body composition, nutritional status, and metabolic biomarkers in PD. Methods CSF cf‐mtDNA levels, defined as the copy numbers of two regions of the mtDNA circular molecule (mt64‐ND1 and mt96‐ND5), were quantified in 44 PD patients and 43 controls using multiplex digital PCR. The mt96‐ND5/mt64‐ND1 ratio was calculated to estimate mtDNA deletion burden. Associations with clinical features, body composition, serum nutritional markers, and plasma energy metabolism‐related organic acids were examined. Generalized linear models (GLMs) were performed to adjust for confounders. Results CSF mt64‐ND1 and mt96‐ND5 levels were lower in PD patients than controls ( p  = 0.002, p  = 0.001), while the mt96‐ND5/mt64‐ND1 ratio showed no group difference. GLM analysis identified body composition indices and serum albumin as key determinants of cf‐mtDNA levels. Subgroup analysis showed lower cf‐mtDNA levels in PD patients with preserved body composition and nutritional status. The mt96‐ND5/mt64‐ND1 ratio displayed a biphasic association with body composition and an inverse correlation with plasma 2‐ketoglutaric acid, suggesting a link to energy metabolism. Interpretation CSF cf‐mtDNA levels are reduced in PD and influenced by body composition and nutritional status, supporting their role as a metabolic biomarker. While the cf‐mtDNA deletion ratio remained unchanged, its association with body composition suggests a complex interplay between mitochondrial integrity and metabolism. These findings highlight the relevance of cf‐mtDNA in PD pathophysiology and the need for further study.