Farmacokinetiek bij Veroudering: Optimalisatie van Farmacologische Therapieën voor Geriatrische Patiënten

The global population is aging, with 11% of Belgium's population in 2022 aged 75 and older. Geriatric patients, often multimorbid and polymedicated, undergo physiological changes that affect drug pharmacokinetics (PK) and pharmacodynamics (PD). Our group identified high PK variability, particularly in drug absorption, which remains unexplained. Changes in gastrointestinal (GI) physiology, such as intraluminal pH and transit times, may contribute to this variability, although data on GI alterations in older adults is lacking. This variability is especially concerning for high-risk medications like oral anticoagulants. Direct oral anticoagulants (DOACs) are preferred over vitamin K antagonists for stroke prevention in geriatric atrial fibrillation (AF) patients, but data on their use in this population is limited. Vancomycin, another high-risk drug due to its narrow therapeutic window, requires therapeutic drug monitoring (TDM). Recent TDM guidelines now focus on AUC24h/MIC, complicating monitoring. Model-informed precision dosing using population pharmacokinetics (popPK) models facilitates vancomycin dosing, but its predictive performance in older patients with high variability remains uncertain. The aim of this PhD thesis is to investigate the pharmacokinetics and GI physiology in geriatric patients and older adults aged 75 years and older, to ultimately improve the safety and efficacy of commonly used, high-risk medications with narrow therapeutic windows in this vulnerable population. In Chapter 1, a prospective study was conducted in geriatric inpatients and community-dwelling adults to assess GI transit and intraluminal pH using a wireless motility capsule system. Data from these participants were combined with those from a previous study involving healthy young volunteers. Multivariable linear regression models were developed to identify factors influencing GI transit time and pH. We recruited 30 subjects (median age 80 [IQR=77-83] years). When compared to young volunteers (median age 21 [20-33] years), geriatric inpatients (N=8) exhibited prolonged gastric emptying times (p<0.01). Community-dwelling older adults (N=22) did not demonstrate delayed gastric emptying (p=0.85). Proton pump inhibitors (PPI) use was associated with increased gastric pH (p<0.01) in the pooled analysis. In Chapter 2, a systematic search was conducted to identify PK/PD studies of apixaban, dabigatran, edoxaban, and rivaroxaban in older adults. The review found that age alone did not significantly impact the exposure of edoxaban, rivaroxaban, or dabigatran, while apixaban peak concentrations were 40% higher in older adults compared to young volunteers. However, considerable interindividual variability in DOAC exposure was observed in older adults. This variability can be attributed to factors such as impaired kidney function, changes in body composition (particularly reduced muscle mass), and the use of P-gp inhibitors, aligning with current dose reduction guidelines for apixaban, edoxaban, and rivaroxaban. Dabigatran exhibited the greatest variability, as its dose adjustment criterion is based solely on age, making it a less preferable option for older adults. Moreover, lower and higher DOAC exposure were strongly associated with stroke and bleeding events, respectively, although no specific therapeutic range for these outcomes have been established. The insights from this review guided the design of the prospective PK study in the following chapter. In the prospective PK study (Chapter 3), edoxaban PK was evaluated in geriatric patients. A population PK (popPK) model was developed using a nonlinear mixed-effects approach to simulate various dosing regimens. The model, based on data from frail geriatric patients (median age 87, IQR 83-90), was best described by a two-compartment structure, with proton pump inhibitors decreasing relative bioavailability. Nearly half of the patients exhibited higher-than-expected trough concentrations, and edoxaban concentrations were strongly correlated with anti-Xa activity levels. Simulations demonstrated that a 15 mg dose had a 72.3% probability of achieving the same range trough concentrations observed in ENGAGE-AF TIMI 48 study, compared to 54.7% for the 30 mg dose and 15.7% for the 60 mg dose. In Chapter 4, we conducted a retrospective analysis of early vancomycin concentrations in older and younger patients. Multivariable analyses were used to evaluate the relationship between baseline characteristics and supratherapeutic exposure (logistic regression) as well as dose-normalized concentrations (linear regression). The study included 449 patients aged ≥75 years (median age 80) and 1,609 patients aged <75 years (median age 61). Univariable analysis revealed that older adults had significantly higher initial vancomycin concentrations (p<0.001) and more frequently reached supratherapeutic levels. In the multivariable analysis, decreased eGFR (CKD-EPI) was associated with supratherapeutic concentrations, while administering lower loading and maintenance doses was inversely related to supratherapeutic concentrations. Predictors of higher dose-normalized concentrations included decreased eGFR (CKD-EPI), lower body weight, elevated blood urea nitrogen, and delayed timing of therapeutic drug monitoring. Finally in Chapter 5, published popPK models were evaluated for their predictive performance using retrospective data from older patients on vancomycin, each with at least three vancomycin concentration measurements. The performance of individual models and a multi-model algorithm—incorporating model selection and averaging—was assessed through (i) a priori predictions based solely on patient covariates and (ii) a posteriori predictions using both covariates and prior vancomycin concentrations. Data from 255 older adults (median age: 79, IQR: 77-84) contributing 765 vancomycin concentrations were analyzed. The most accurate dosing predictions in this population were achieved when using the most recent sample, with popPK models developed specifically for older populations performing best. The Thomson model demonstrated the highest accuracy, followed closely by the Ji model. In conclusion, edoxaban pharmacokinetics were significantly altered in frail geriatric patients compared to those observed in pivotal trials, with lower clearance, reduced volume of distribution, and a slower absorption rate. The delayed gastric emptying time in frail elderly patients likely contributed to the slower absorption. Advanced age did not directly influence PK. The effect of age was confounded by covariates, such as declining renal and liver function, as shown in the vancomycin study. These changes can be effectively captured and modeled using popPK approaches.