Bayesian clinical decision support-guided versus clinician-guided vancomycin dosing in attainment of targeted pharmacokinetic parameters in a paediatric population.

OBJECTIVES: To compare a Bayesian clinical decision support (CDS) dose-optimizing software program with clinician judgement in individualizing vancomycin dosing regimens to achieve vancomycin pharmacokinetic (PK)/pharmacodynamic (PD) targets in a paediatric population. METHODS: A retrospective review combined with a model-based simulation of vancomycin dosing was performed on children aged 1 year to 18 years at the University of California, San Francisco Benioff Children's Hospital Mission Bay. Dosing regimens recommended by the clinical pharmacists, 'clinician-guided', were compared with alternative 'CDS-guided' dosing regimens. The primary outcome was the percentage of occasions predicted to achieve steady-state trough levels within the target range of 10-15 mg/L, with a secondary outcome of predicted attainment of AUC24 ≥400 mg·h/L. Statistical comparison between approaches was performed using a standard t-test. RESULTS: A total of n=144 patient occasions were included. CDS-guided regimens were predicted to achieve vancomycin steady-state troughs in the target range on 70.8% (102/144) of occasions, as compared with 37.5% (54/144) in the clinician-guided arm (P<0.0001). An AUC24 of ≥400 mg·h/L was achieved on 93% (112/121) of occasions in the CDS-guided arm versus 72% (87/121) of occasions in the clinician-guided arm (P<0.0001). CONCLUSIONS: In a simulated analysis, the use of a Bayesian CDS tool was better than clinician judgement in recommending vancomycin dosing regimens in which PK/PD targets would be attained in children.

Clinical Decision Support for Precision Dosing: Opportunities for Enhanced Equity and Inclusion in Health Care.

Precision dosing aims to tailor doses to individual patients with the goal of improving treatment efficacy and avoiding toxicity. Clinical decision support software (CDSS) plays a crucial role in mediating this process, translating knowledge derived from clinical trials and real-world data (RWD) into actionable insights for clinicians to use at the point of care. However, not all patient populations are proportionally represented in clinical trials and other data sources that inform CDSS tools, limiting the applicability of these tools for underrepresented populations. Here, we review some of the limitations of existing CDSS tools and discuss methods for overcoming these gaps. We discuss considerations for study design and modeling to create more inclusive CDSS, particularly with an eye toward better incorporation of biological indicators in place of race, ethnicity, or sex. We also review inclusive practices for collection of these demographic data, during both study design and in software user interface design. Because of the role CDSS plays in both recording routine clinical care data and disseminating knowledge derived from data, CDSS presents a promising opportunity to continuously improve precision dosing algorithms using RWD to better reflect the diversity of patient populations.

Clinical decision support for chemotherapy-induced neutropenia using a hybrid pharmacodynamic/machine learning model.

Consensus guidelines recommend use of granulocyte colony stimulating factor in patients deemed at risk of chemotherapy-induced neutropenia, however, these risk models are limited in the factors they consider and miss some cases of neutropenia. Clinical decision making could be supported using models that better tailor their predictions to the individual patient using the wealth of data available in electronic health records (EHRs). Here, we present a hybrid pharmacokinetic/pharmacodynamic (PKPD)/machine learning (ML) approach that uses predictions and individual Bayesian parameter estimates from a PKPD model to enrich an ML model built on her data. We demonstrate this approach using models developed on a large real-world data set of 9121 patients treated for lymphoma, breast, or thoracic cancer. We also investigate the benefits of augmenting the training data using synthetic data simulated with the PKPD model. We find that PKPD-enrichment of ML models improves prediction of grade 3-4 neutropenia, as measured by higher precision (61%) and recall (39%) compared to PKPD model predictions (47%, 33%) or base ML model predictions (51%, 31%). PKPD augmentation of ML models showed minor improvements in recall (44%) but not precision (56%), and data augmentation required careful tuning to control overfitting its predictions to the PKPD model. PKPD enrichment of ML shows promise for leveraging both the physiology-informed predictions of PKPD and the ability of ML to learn predictor-outcome relationships from large data sets to predict patient response to drugs in a clinical precision dosing context.

The Relationship Between Busulfan Exposure and Achievement of Sustained Donor Myeloid Chimerism in Patients with Non-Malignant Disorders.

The overall objective of allogeneic hematopoietic cell transplantation (HCT) in patients with non-malignant conditions involves replacing a dysfunctional or absent cell or gene product for disease correction. It is unclear whether lower busulfan exposure may be sufficient in this population to facilitate durable myeloid engraftment and limit toxicity. Given that neither the ideal level of mixed myeloid chimerism for specific non-malignant diseases nor how to condition a patient to achieve stable mixed myeloid chimerism is fully known, we sought to analyze the relationships among busulfan exposure, myeloid chimerism, and outcomes in patients with non-malignant conditions receiving busulfan as a part of combination pretransplant conditioning at our institution. This was a single-center, retrospective study including pediatric patients with a variety of non-malignant disorders who underwent allogeneic HCT at the University of California San Francisco Benioff Children's Hospital from March 2007 to June 2018. The busulfan cumulative area under the curve (cAUC) was estimated using a validated population pharmacokinetic model and nonlinear mixed effects modeling. Median busulfan cAUC for all patients was 70 mg·h/L (range, 53 to 108). All of the 29 patients with a busulfan cAUC of ≥70 mg·h/L achieved long-term disease correction with full or stable mixed (>20%) myeloid chimerism, compared to 78.5% (22/28) of patients with a cAUC of <70 mg·h/L (P = .01). Overall ksurvival was evaluated up to 3 years and was identical in patients with busulfan cAUC < 70 mg·h/L and patients with busulfan cAUC ≥70 mg·h/L (96% versus 93%; P = .92). Only three patients died, at days 65, 164 and 980 days post-HCT. Severe busulfan-related toxicities and graft-versus-host-disease (GVHD) were rare, with veno-occlusive disease occurring in four patients (7%), acute respiratory distress syndrome in three patients (5%), and GVHD in five patients (9%). These results demonstrate excellent outcomes and extremely low rates of toxicity across our entire cohort. Based on the results of this study, we recommend a busulfan exposure target of 75 mg·h/L (range, 70 to 80) in all non-malignant patients receiving allogeneic HCT to ensure optimal exposure for achievement of high-level stable myeloid chimerism.

Continuous Learning in Model-Informed Precision Dosing: A Case Study in Pediatric Dosing of Vancomycin.

Model-informed precision dosing (MIPD) leverages pharmacokinetic (PK) models to tailor dosing to an individual patient's needs, improving attainment of therapeutic drug exposure targets and thus potentially improving drug efficacy or reducing adverse events. However, selection of an appropriate model for supporting clinical decision making is not trivial. Error or bias in dose selection may arise if the selected model was developed in a population not fully representative of the intended MIPD population. One previously proposed approach is continuous learning, in which an initial model is used in MIPD and then updated as additional data becomes available. In this case study of pediatric vancomycin MIPD, the potential benefits of the continuous learning approach are investigated. Five previously published models were evaluated and found to perform adequately in a data set of 273 pediatric patients in the intensive care unit. Additionally, two predefined simple PK models were fitted on separate populations of 50-350 patients in an approach mimicking clinical implementation of automated continuous learning. With these continuous learning models, prediction error using population PK parameters could be reduced by 2-13% compared with previously published models. Sample sizes of at least 200 patients were found suitable for capturing the interindividual variability in vancomycin at this institution, with limited benefits of larger data sets. Although comprised mostly of trough samples, these sparsely sampled routine clinical data allowed for reasonable estimation of simulated area under the curve (AUC). Together, these findings lay the foundations for a continuous learning MIPD approach.

The Pharmacokinetic and Pharmacodynamic Properties of Hydroxychloroquine and Dose Selection for COVID-19: Putting the Cart Before the Horse.

Coronavirus disease 2019 (COVID-19), caused by the 2019 novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is currently responsible for a global pandemic. To date, only remdesivir and dexamethasone have demonstrated a positive response in a prospective, randomized trial for the treatment of patients with COVID-19. Hydroxychloroquine (HCQ) is an agent available in an oral formulation with in vitro activity against SARS-CoV-2 that has been suggested as a potential agent. Unfortunately, results of randomized trials evaluating HCQ as treatment against a control group are lacking, and little is known about its pharmacokinetic/pharmacodynamic (PK/PD) profile against SARS-CoV-2. The objective of this review was to describe the current understanding of the PK/PD and dose selection of HCQ against SARS-CoV-2, discuss knowledge gaps, and identify future studies that are needed to optimize the efficacy and safety of treatments against COVID-19.

Assessment of a Model-Informed Precision Dosing Platform Use in Routine Clinical Care for Personalized Busulfan Therapy in the Pediatric Hematopoietic Cell Transplantation (HCT) Population.

INTRODUCTION: Population pharmacokinetic (PK) studies demonstrate model-based dosing for busulfan that incorporates body size and age improve clinical target attainment as compared to weight-based regimens. Recently, for clinical dosing of busulfan and TDM, our institution transitioned to a cloud-based clinical decision support tool (www.insight-rx.com). The goal of this study was to assess the dose decision tool for the achievement of target exposure of busulfan in children undergoing hematopoietic cell transplantation (HCT). PATIENTS AND METHODS: Patients (N = 188) were grouped into cohorts A, B, or C based on the method for initial dose calculation and estimation of AUC: Cohort A: Initial doses were based on the conventional dosing algorithm (as outlined in the manufacturers' package insert) and non-compartmental analysis (NCA) estimation using the trapezoidal rule for estimation of AUC following TDM. Cohort B: Initial doses for busulfan were estimated by a first-generation PK model and NCA estimation of AUC following TDM. Cohort C: Initial doses were calculated by an updated, second-generation PK model available in the dose decision tool with an estimation of AUC following TDM. RESULTS: The percent of individuals achieving the exposure target at the time of first PK collection was higher in subjects receiving initial doses provided by the model-informed precision dosing platform (cohort C, 75%) versus subjects receiving initial doses based on either of the two other approaches (conventional guidelines/cohort A, 25%; previous population PK model and NCA parameter estimation, cohort B, 50%). Similarly, the percent of subjects achieving the targeted cumulative busulfan exposure (cAUC) in cohort C was 100% vs. 66% and 88% for cohort A and B, respectively. For cAUC, the variability in the spread of target attainment (%CV) was low at 4.1% for cohort C as compared to cohort A (14.8%) and cohort B (17.1%). CONCLUSION: Achievement of goal exposure early on in treatment was improved with the updated model for busulfan and the Bayesian platform. Model-informed dosing and TDM utilizing a Bayesian-based platform provides a significant advantage over conventional guidelines for the achievement of goal cAUC exposure.

Model-Informed Precision Dosing of Vancomycin in Hospitalized Children: Implementation and Adoption at an Academic Children's Hospital.

BACKGROUND: Model-informed precision dosing (MIPD) can serve as a powerful tool during therapeutic drug monitoring (TDM) to help individualize dosing in populations with large pharmacokinetic variation. Yet, adoption of MIPD in the clinical setting has been limited. Overcoming technologic hurdles that allow access to MIPD at the point-of-care and placing it in the hands of clinical specialists focused on medication dosing may encourage adoption. OBJECTIVE: To describe the hospital implementation and usage of a MIPD clinical decision support (CDS) tool for vancomycin in a pediatric population. METHODS: Within an academic children's hospital, MIPD for vancomycin was implemented via a commercial cloud-based CDS tool that utilized Bayesian forecasting. Clinical pharmacists were recognized as local champions to facilitate adoption of the tool and operated as end-users. Integration within the electronic health record (EHR) and automatic transmission of patient data to the tool were identified as important requirements. A web-link icon was developed within the EHR which when clicked sends users and needed patient-level clinical data to the CDS platform. Individualized pharmacokinetic predictions and exposure metrics for vancomycin are then presented in the form of a web-based dashboard. Use of the CDS tool as part of TDM was tracked and users were surveyed on their experience. RESULTS: After a successful pilot phase in the neonatal intensive care unit, implementation of MIPD was expanded to the pediatric intensive care unit, followed by availability to the entire hospital. During the first 2+ years since implementation, a total of 853 patient-courses (n = 96 neonates, n = 757 children) and 2,148 TDM levels were evaluated using the CDS tool. For the most recent 6 months, the CDS tool was utilized to support 79% (181/230) of patient-courses in which TDM was performed. Of 26 users surveyed, > 96% agreed or strongly agreed that automatic transmission of patient data to the tool was a feature that helped them complete tasks more efficiently; 81% agreed or strongly agreed that they were satisfied with the CDS tool. CONCLUSIONS: Integration of a vancomycin CDS tool within the EHR, along with leveraging the expertise of clinical pharmacists, allowed for successful adoption of MIPD in clinical care.

Individualized Empiric Vancomycin Dosing in Neonates Using a Model-Based Approach.

BACKGROUND: Vancomycin dosing in neonates is challenging because of the large variation in pharmacokinetics. Existing empiric dosing recommendations use table-based formats, within which a neonate is categorized on the basis of underlying characteristics. The ability to individualize dosing is limited because of the small number of "dose categories," and achieving narrow exposure targets is difficult. Our objective was to evaluate a model-based dosing approach (which we designated Neo-Vanco) designed to individualize empiric vancomycin dosing in neonates. METHODS: Neo-Vanco was developed on the basis of a published, externally validated population pharmacokinetic model. Using a simulation-based methodology, individualized empiric doses that maximize the probability of attaining a 24-hour area under the curve/minimum inhibitory concentration ratio (AUC24/MIC) of >400 while minimizing troughs >20 mg/L are calculated. To evaluate the Neo-Vanco strategy, retrospective data from neonates treated with vancomycin at 2 healthcare systems were used, and empiric dose recommendations from the following 4 sources were examined: Neo-Vanco, Neofax, Red Book, and Lexicomp. Predicted AUC24 and troughs were calculated and compared. RESULTS: Overall, 492 neonates were evaluated (median postmenstrual age, 36 weeks [5th-95th percentiles (90% range), 25-47 weeks]; median weight, 2.4 kg [90% range, 0.6-4.8 kg]). The percentage of neonates predicted to achieve an AUC24/MIC of >400 was 94% with Neo-Vanco, 18% with Neofax, 23% with Red Book, and 55% with Lexicomp (all P < .0001 vs Neo-Vanco). Predicted troughs of >20 mg/L were infrequent and similar across the dosing approaches (Neo-Vanco, 2.8%; Neofax, 1.0% [P = .03]; Red Book, 2.6% [P = .99]; and Lexicomp, 4.1% [P = .27]. CONCLUSION: A model-based dosing approach that individualizes empiric vancomycin dosing was predicted to improve achievement of target exposure levels in neonates. Prospective clinical evaluation is warranted.

Model-Informed Precision Dosing at the Bedside: Scientific Challenges and Opportunities.

The development of model-informed precision dosing (MIPD) tools, especially in the form of native or web-based applications to be used at the bedside, has garnered marked attention in recent years. Their potential clinical benefit can be large, but it should be ensured that such tools make optimal use of available clinical data and have adequate predictive ability. Unique scientific challenges specific to MIPD remain, which may require adaptation of commonly used diagnostics in pharmacometrics.

The Effect of Famotidine, a MATE1-Selective Inhibitor, on the Pharmacokinetics and Pharmacodynamics of Metformin.

INTRODUCTION: Pharmacokinetic outcomes of transporter-mediated drug-drug interactions (TMDDIs) are increasingly being evaluated clinically. The goal of our study was to determine the effects of selective inhibition of multidrug and toxin extrusion protein 1 (MATE1), using famotidine, on the pharmacokinetics and pharmacodynamics of metformin in healthy volunteers. METHODS: Volunteers received metformin alone or with famotidine in a crossover design. As a positive control, the longitudinal effects of famotidine on the plasma levels of creatinine (an endogenous substrate of MATE1) were quantified in parallel. Famotidine unbound concentrations in plasma reached 1 µM, thus exceeding the in vitro concentrations that inhibit MATE1 [concentration of drug producing 50 % inhibition (IC50) 0.25 µM]. Based on current regulatory guidance, these concentrations are expected to inhibit MATE1 clinically [i.e. maximum unbound plasma drug concentration (C max,u)/IC50 >0.1]. RESULTS: Consistent with MATE1 inhibition, famotidine administration significantly altered creatinine plasma and urine levels in opposing directions (p < 0.005). Interestingly, famotidine increased the estimated bioavailability of metformin [cumulative amount of unchanged drug excreted in urine from time zero to infinity (A e∞)/dose; p < 0.005] without affecting its systemic exposure [area under the plasma concentration-time curve (AUC) or maximum concentration in plasma (C max)] as a result of a counteracting increase in metformin renal clearance. Moreover, metformin-famotidine co-therapy caused a transient effect on oral glucose tolerance tests [area under the glucose plasma concentration-time curve between time zero and 0.5 h (AUCglu,0.5); p < 0.005]. CONCLUSIONS: These results suggest that famotidine may improve the bioavailability and enhance the renal clearance of metformin.