Machine-Learning Assisted Screening of Correlated Covariates: Application to Clinical Data of Desipramine.

Stepwise covariate modeling (SCM) has a high computational burden and can select the wrong covariates. Machine learning (ML) has been proposed as a screening tool to improve the efficiency of covariate selection, but little is known about how to apply ML on actual clinical data. First, we simulated datasets based on clinical data to compare the performance of various ML and traditional pharmacometrics (PMX) techniques with and without accounting for highly-correlated covariates. This simulation step identified the ML algorithm and the number of top covariates to select when using the actual clinical data. A previously developed desipramine population-pharmacokinetic model was used to simulate virtual subjects. Fifteen covariates were considered with four having an effect included. Based on the F1 score (an accuracy measure), ridge regression was the most accurate ML technique on 200 simulated datasets (F1 score = 0.475 ± 0.231), a performance which almost doubled when highly-correlated covariates were accounted for (F1 score = 0.860 ± 0.158). These performances were better than forwards selection with SCM (F1 score = 0.251 ± 0.274 and 0.499 ± 0.381 without/with correlations respectively). In terms of computational cost, ridge regression (0.42 ± 0.07 seconds/simulated dataset, 1 thread) was ~20,000 times faster than SCM (2.30 ± 2.29 hours, 15 threads). On the clinical dataset, prescreening with the selected ML algorithm reduced SCM runtime by 42.86% (from 1.75 to 1.00 days) and produced the same final model as SCM only. In conclusion, we have demonstrated that accounting for highly-correlated covariates improves ML prescreening accuracy. The choice of ML method and the proportion of important covariates (unknown a priori) can be guided by simulations.

Needs-driven talent and competency development for the next generation of regulatory scientists in Africa.

Capacity building programmes for African regulators should link education, training and research with career development in an approach that combines an academic base and experiential learning aligned within a competency framework. A regulatory ecosystem that engages with a broad range of stakeholders will mean that expertise in the ever-expanding field of regulatory science filters into teaching and research in a symbiotic way. In this way capacity development interventions will be a collaborative approach between the learning context (academic and training institutions) and the performance context (regulatory agencies and industry), which will ultimately best serve the patients. Monitoring and evaluation of capacity development interventions will be essential to show value of investments and ultimately guide continued funding and sustainability. This paper reviews the skills and human capacity gaps, reports on regulatory assessment pathways used in Ghana, South Africa and Zimbabwe and outlines a staged tactical approach for Africa that builds on previous efforts to strengthen African regulatory ecosystems.

Emerging Therapeutics, Technologies, and Drug Development Strategies to Address Patient Nonadherence and Improve Tuberculosis Treatment.

Imperfect medication adherence remains the biggest predictor of treatment failure for patients with tuberculosis. Missed doses during treatment lead to relapse, tuberculosis resistance, and further spread of disease. Understanding individual patient phenotypes, population pharmacokinetics, resistance development, drug distribution to tuberculosis lesions, and pharmacodynamics at the site of infection is necessary to fully measure the impact of adherence on patient outcomes. To decrease the impact of expected variabilityin drug intake on tuberculosis outcomes, an improvement in patient adherence and new forgiving regimens that protect against missed doses are needed. In this review, we summarize emerging technologies to improve medication adherence in clinical practice and provide suggestions on how digital adherence technologies can be incorporated in clinical trials and practice and the drug development pipeline that will lead to more forgiving regimens and benefit patients suffering from tuberculosis.

Traditional medicines for COVID-19: Perspectives from clinical pharmacologists.

There has been high interest in the use of traditional medicines for COVID-19 from early in the course of the pandemic. Significant advances in the science of ethnopharmacology have helped to introduce chemical entities identified from natural sources into modern medicine. However, the wider integration of natural products into the modern drug discovery process will require enhanced collaboration amongst the pharmaceutical industry, academic research units, regulatory bodies, ethics review committees and local, regional, continental and international organizations. Revisiting this topic holds promise of benefit for both the current and future pandemics.

Clinical Pharmacology Worldwide: A Global Health Perspective.

Low- and middle-income countries (LMICs) have the highest rates of mortality and morbidity globally, but lag behind high-income countries in the number of clinical trials and trained researchers, as well as research data pertaining to their populations. Lack of local clinical pharmacology and pharmacometrics expertise, limited training opportunities, and lack of local genomic data may contribute to health inequalities and limit the application of precision medicine. Continuing to develop health care infrastructure, including well-designed clinical pharmacology training and data collection in LMICs, can help address these challenges. International collaboration aimed at improving training and infrastructure and encouraging locally driven research and clinical trials will be of benefit. This review describes several examples where clinical pharmacology expertise could be leveraged, including opportunities for pharmacogenomic expertise that could drive improved recommendations for clinical guidelines. Also described are clinical pharmacology and pharmacometrics training programs in Africa, and the personal experience of a Tanzanian researcher currently on a training sabbatical in the United States, as illustrative examples of how training in clinical pharmacology can be effectively implemented in LMICs. These training efforts will benefit from advocacy for employment opportunities and career development pathways for clinical pharmacologists that are gradually being recognized and developed in LMICs. Clinical pharmacologists have a key role to play in global health, and development of training and research infrastructure to advance this expertise in LMICs will be of tremendous benefit.

Dolutegravir pharmacokinetics during co-administration with either artemether/lumefantrine or artesunate/amodiaquine.

BACKGROUND: In sub-Saharan Africa, artemisinin-containing therapies for malaria treatment are regularly co-administered with ART. Currently, dolutegravir-based regimens are recommended as first-line therapy for HIV across most of Africa. OBJECTIVES: To investigate the population pharmacokinetics of dolutegravir during co-administration with artemether/lumefantrine or artesunate/amodiaquine, two commonly used antimalarial therapies. METHODS: We developed a population pharmacokinetic model of dolutegravir with data from 26 healthy volunteers in two Phase 2 studies with a total of 403 dolutegravir plasma concentrations at steady state. Volunteers received 50 mg of dolutegravir once daily alone or in combination with standard treatment doses of artemether/lumefantrine (80/480 mg) or artesunate/amodiaquine (200/540 mg). RESULTS: A two-compartment model with first-order elimination and transit compartment absorption best described the concentration-time data of dolutegravir. Typical population estimates for clearance, absorption rate constant, central volume, peripheral volume and mean absorption transit time were 0.713 L/h, 1.68 h-1, 13.2 L, 5.73 L and 1.18 h, respectively. Co-administration of artemether/lumefantrine or artesunate/amodiaquine increased dolutegravir clearance by 10.6% (95% CI 4.09%-34.5%) and 26.4% (95% CI 14.3%-51.4%), respectively. Simulations showed that simulated trough concentrations of dolutegravir alone or in combination with artemether/lumefantrine or artesunate/amodiaquine are maintained above the dolutegravir protein-adjusted IC90 of 0.064 mg/L for more than 99% of the individuals. CONCLUSIONS: Dolutegravir dose adjustments are not necessary for patients who are taking standard 3 day treatment doses of artemether/lumefantrine or artesunate/amodiaquine.

Risk reduction of diarrhea and respiratory infections following a community health education program - a facility-based case-control study in rural parts of Kenya.

BACKGROUND: Diarrheal and acute respiratory infections remain a major cause of death in developing countries especially among children below 5 years of age. About 80% of all hospital attendances in Kenya can be attributed to preventable diseases and at least 50% of these preventable diseases are linked to poor sanitation. The purpose of this study was to assess the impact of a community-based health education program, called Familia Nawiri, in reducing the risk of diarrhea and respiratory infections among people living in three rural Kenyan communities. METHODS: Cases were defined as patients attending the health facility due to diarrhea or a respiratory infection while controls were patients attending the same health facility for a non-communicable disease defined as an event other than diarrhea, respiratory infection. Adjusted odds ratios (ORs) with 95% confidence intervals (CIs) were calculated using a logistic regression model to assess the risk of diarrheal or respiratory infection in association with exposure to the health education program. RESULTS: There were 324 cases and 308 controls recruited for the study with 57% of the cases and 59% of the controls being male. Overall, 13% of cases vs. 20% of control patients were exposed to the education program. Participants exposed to the program had 38% lower odds of diarrhea and respiratory infections compared to those not exposed to the program (adjusted OR 0.62, 95% CI 0.41-0.96). A similar risk reduction was observed for participants in the study who resided in areas with water improvement initiatives (adjusted OR 0.65, 95% CI 0.47-0.90). Variables in the adjusted model included water improvement projects in the area and toilet facilities. CONCLUSION: Findings from this study suggest participants exposed to the education program and those residing in areas with water improvement initiatives have a reduced risk of having diarrhea or respiratory infection.

Care-Seeking Dynamics among Patients with Diabetes Mellitus and Hypertension in Selected Rural Settings in Kenya.

Diabetes mellitus and hypertension are two common non-communicable diseases (NCDs) that often coexist in patients. However, health-seeking behaviour in patients with diabetes mellitus or hypertension has not been extensively studied especially in low- and middle-income countries. This study aimed to examine care-seeking dynamics among participants diagnosed with diabetes and/or hypertension across nine counties in rural Kenya. We conducted a cross-sectional study among adults diagnosed with diabetes and/or hypertension through face-to-face interviews. Of the 1100 participants, 69.9% had hypertension, 15.5% diabetes while 14.7% had both. The mean age of the respondents was 64 years. The majority of the respondents (86%) were on allopathic treatment. Hospital admission, having a good self-rated health status and having social support for illness, were positively associated with appropriate health-seeking behaviour while use of alcohol and pharmacy or chemist as source of treatment were negatively associated with appropriate health-seeking behaviour. Our study found a high prevalence of appropriate health-seeking behaviour among respondents with the majority obtaining care from government facilities. The results are evidence that improving public health care services can promote appropriate health-seeking behaviour for non-communicable diseases and thus improve health outcomes.

The Next Generation Scientist program: capacity-building for future scientific leaders in low- and middle-income countries.

BACKGROUND: Scientific and professional development opportunities for early career scientists in low- and middle- income countries (LMICs) are limited and not consistent. There is a disproportionately low number of biomedical and clinical researchers in LMIC's relative to their high burden of disease, a disparity that is aggravated by emigration of up to 70% of scientists from their countries of birth for education and employment elsewhere. To help address this need, a novel University-accredited, immersive fellowship program was established by a large public-academic-private network. We sought to describe the program and summarize progress and lessons learned over its first 7-years. METHODS: Hallmarks of the program are a structured learning curriculum and bespoke research activities tailored to the needs of each fellow. Research projects expose the scientists to state-of-the-art methodologies and leading experts in their fields while also ensuring that learnings are implementable within their home infrastructure. Fellows run seminars on drug discovery and development that reinforce themes of scientific leadership and teamwork together with practical modules on addressing healthcare challenges within their local systems. Industry mentors achieve mutual learning to better understand healthcare needs in traditionally underserved settings. We evaluated the impact of the program through an online survey of participants and by assessing research output. RESULTS: More than 140 scientists and clinicians from 25 countries participated over the 7-year period. Evaluation revealed strong evidence of knowledge and skills transfer, and beneficial self-reported impact on fellow's research output and career trajectories. Examples of program impact included completion of post-graduate qualifications; establishment and implementation of good laboratory- and clinical- practice mechanisms; and becoming lead investigators in local programs. There was a high retention of fellows in their home countries (> 75%) and an enduring professional network among the fellows and their mentors. CONCLUSIONS: Our experience demonstrates an example for how multi-sectoral partners can contribute to scientific and professional development of researchers in LMICs and supports the idea that capacity-building efforts should be tailored to the specific needs of beneficiaries to be maximally effective. Lessons learned may be applied to the design and conduct of other programs to strengthen science ecosystems in LMICs.

Open-source mobile digital platform for clinical trial data collection in low-resource settings.

BACKGROUND: Governments, universities and pan-African research networks are building durable infrastructure and capabilities for biomedical research in Africa. This offers the opportunity to adopt from the outset innovative approaches and technologies that would be challenging to retrofit into fully established research infrastructures such as those regularly found in high-income countries. In this context we piloted the use of a novel mobile digital health platform, designed specifically for low-resource environments, to support high-quality data collection in a clinical research study. OBJECTIVE: Our primary aim was to assess the feasibility of a using a mobile digital platform for clinical trial data collection in a low-resource setting. Secondarily, we sought to explore the potential benefits of such an approach. METHODS: The investigative site was a research institute in Nairobi, Kenya. We integrated an open-source platform for mobile data collection commonly used in the developing world with an open-source, standard platform for electronic data capture in clinical trials. The integration was developed using common data standards (Clinical Data Interchange Standards Consortium (CDISC) Operational Data Model), maximising the potential to extend the approach to other platforms. The system was deployed in a pharmacokinetic study involving healthy human volunteers. RESULTS: The electronic data collection platform successfully supported conduct of the study. Multidisciplinary users reported high levels of satisfaction with the mobile application and highlighted substantial advantages when compared with traditional paper record systems. The new system also demonstrated a potential for expediting data quality review. DISCUSSION AND CONCLUSIONS: This pilot study demonstrated the feasibility of using a mobile digital platform for clinical research data collection in low-resource settings. Sustainable scientific capabilities and infrastructure are essential to attract and support clinical research studies. Since many research structures in Africa are being developed anew, stakeholders should consider implementing innovative technologies and approaches.

Glibenclamide population pharmacokinetic/pharmacodynamic modeling in South African type 2 diabetic subjects.

AIM: To determine the effective dose of glibenclamide by quantifying the dose-response relationship in South African type 2 diabetic patients. PATIENTS AND METHODS: A total of 24 type 2 diabetic patients participated in a glibenclamide dose-escalation study during which glibenclamide, glucose, and insulin concentrations were quantified, while the dose of glibenclamide was progressively increased. All except four subjects contributed data on all dose-escalation steps; however, data from all 24 patients were included in the model-based analysis. Pharmacokinetic/pharmacodynamic (PKPD) relationships were modeled using the software Nonmem®. Six models were utilized to explore the effect of alternative glibenclamide dose and plasma concentration inputs on various metrics of glucose response. RESULTS: Six models adequately described the experimental data. The effective dose for a glucose-lowering effect suggested by PKPD modeling is less than 5 mg/day. Doses beyond 5 mg/day do not meaningfully add to glibenclamide effects on blood-glucose response. CONCLUSION: The effective dose of glibenclamide, suggested by PKPD modeling, is less than 5 mg/day. Higher doses of glibenclamide, eg, 15 mg/day as originally recommended by the manufacturer, do not produce further decrease in the blood glucose level but may predispose the patients to adverse effects.

Population pharmacokinetic modeling of glibenclamide in poorly controlled South African type 2 diabetic subjects.

AIM: The aim of this study was to describe the pharmacokinetics (PK) of glibenclamide in poorly controlled South African type 2 diabetic subjects using noncompartmental and model-based methods. METHODS: A total of 24 subjects with type 2 diabetes were administered increasing doses (0 mg/d, 2.5 mg/d, 5 mg/d, 10 mg/d, and 20 mg/d) of glibenclamide daily at 2-week intervals. Plasma glibenclamide, glucose, and insulin determinations were performed. Blood sampling times were 0 minute, 30 minutes, 60 minutes, 90 minutes, and 120 minutes (post breakfast sampling) and 240 minutes, 270 minutes, 300 minutes, 330 minutes, 360 minutes, and 420 minutes (post lunch sampling) on days 14, 28, 42, 56, and 70 for doses of 0 mg, 2.5 mg, 5.0 mg, 10 mg, and 20 mg, respectively. Blood sampling was performed after the steady state was reached. A total of 24 individuals in the data set contributed to a total of 841 observation records. The PK was analyzed using noncompartmental analysis methods, which were implemented in WinNonLin(®), and population PK analysis using NONMEM(®). Glibenclamide concentration data were log transformed prior to fitting. RESULTS: A two-compartmental disposition model was selected after evaluating one-, two-, and three-compartmental models to describe the time course of glibenclamide plasma concentration data. The one-compartment model adequately described the data; however, the two-compartment model provided a better fit. The three-compartment model failed to achieve successful convergence. A more complex model, to account for enterohepatic recirculation that was observed in the data, was unsuccessful. CONCLUSION: In South African diabetic subjects, glibenclamide demonstrates linear PK and was best described by a two-compartmental model. Except for the absorption rate constant, the other PK parameters reported in this study are comparable to those reported in the scientific literature. The study is limited by the small study sample size and inclusion of poorly controlled type 2 diabetic subjects.

Pharmacokinetic evaluation of the penetration of antituberculosis agents in rabbit pulmonary lesions.

Standard antituberculosis (anti-TB) therapy requires the use of multiple drugs for a minimum of 6 months, with variable outcomes that are influenced by a number of microbiological, pathological, and clinical factors. This is despite the availability of antibiotics that have good activity against Mycobacterium tuberculosis in vitro and favorable pharmacokinetic profiles in plasma. However, little is known about the distribution of widely used antituberculous agents in the pulmonary lesions where the pathogen resides. The rabbit model of TB infection was used to explore the hypothesis that standard drugs have various abilities to penetrate lung tissue and lesions and that adequate drug levels are not consistently reached at the site of infection. Using noncompartmental and population pharmacokinetic approaches, we modeled the rate and extent of distribution of isoniazid, rifampin, pyrazinamide, and moxifloxacin in rabbit lung and lesions. Moxifloxacin reproducibly showed favorable partitioning into lung and granulomas, while the exposure of isoniazid, rifampin, and pyrazinamide in lesions was markedly lower than in plasma. The extent of penetration in lung and lesions followed different trends for each drug. All four agents distributed rapidly from plasma to tissue with equilibration half-lives of less than 1 min to an hour. The models adequately described the plasma concentrations and reasonably captured actual lesion concentrations. Though further refinement is needed to accurately predict the behavior of these drugs in human subjects, our results enable the integration of lesion-specific pharmacokinetic-pharmacodynamic (PK-PD) indices in clinical trial simulations and in in vitro PK-PD studies with M. tuberculosis.

Variability in the population pharmacokinetics of isoniazid in South African tuberculosis patients.

AIM: This study was designed to characterize the population pharmacokinetics of isoniazid in South African pulmonary tuberculosis patients. METHODS: Concentration-time measurements obtained from 235 patients receiving oral doses of isoniazid as part of routine tuberculosis chemotherapy in two clinical studies were pooled and subjected to nonlinear mixed-effects analysis. RESULTS: A two-compartmental model, including first-order absorption and elimination with allometric scaling, was found to describe the observed dose-exposure relationship for oral isoniazid adequately. A mixture model was used to characterize dual rates of isoniazid elimination. Estimates of apparent clearance in slow and fast eliminators were 9.70 and 21.6 l h(-1) , respectively. The proportion of fast eliminators in the population was estimated to be 13.2%. Central volume of distribution was estimated to be 10% smaller in female patients and clearance was found to be 17% lower in patients with HIV. Variability in absorption rate (90%) was completely interoccasional in nature, whereas in relative bioavailability, interoccasional variability (8.4%) was lower than interindividual variability (26%). Oral doses, given once daily according to dosing policies at the time, were sufficient to reach therapeutic concentrations in the majority of the studied population, regardless of eliminator phenotype. Simulations suggested that current treatment guidelines (5 mg kg(-1) ) may be suboptimal in fast eliminators with low body weight. CONCLUSIONS: A population pharmacokinetic model was developed to characterize the highly variable pharmacokinetics of isoniazid in a South African pulmonary tuberculosis patient population. Current treatment guidelines may lead to underexposure in rapid isoniazid eliminators.

Population pharmacodynamic model of the longitudinal FEV1 response to an inhaled long-acting anti-muscarinic in COPD patients.

The precise assessment of the dose-response to bronchodilators in the treatment of chronic obstructive pulmonary disease is hindered by the low signal to noise ratio of the typical clinical endpoint FEV(1). Kinetic-pharmacodynamic (K-PD) models which use time course of response over a range of doses are in principle suited for the assessment of the dose response relationship of pulmonary administered drugs. A K-PD model was successfully developed using the longitudinal FEV1 data collected in the clinical study for a novel bronchodilator X. A superposition of two cosine functions was selected to describe the circadian variability in FEV(1) at baseline. The onset (ka) and offset (kde) of drug action were described with first-order rate constants of 0.214/h and 0.141/h, respectively. Drug potency, EKD(50,) was estimated as 6.56 µg/h, and the maximum response, Emax as 0.631 L. Between-subject variability for kde, EKD(50) and Emax were 65, 84.7 or 34.4% (expressed as coefficient variation). The model-based simulation predicted that for the same total daily dose of once-daily and twice-daily regimens, the trough FEV(1) response to a twice-daily regimen was higher, and the maximum FEV(1) response to once-daily regimen was higher, while the predicted average FEV(1) response was about the same.

Population pharmacokinetics of valsartan in pediatrics.

The objective of this work was to develop a population pharmacokinetic model to assess the influence of subject covariates on the pharmacokinetics of valsartan in children. Data were collected from a single dose study in 26 hypertensive children ages 1 to 16 years. Subjects received 2 mg/kg valsartan suspension up to a maximum dose of 80 mg. Plasma samples were collected and analyzed using LC/MS/MS. Several structural pharmacokinetic models were evaluated for appropriateness. Allometric scaling and standard covariate analyses were performed to explain interindividual variabilities. Objective function values and goodness of fit plots were used for model selection. A posterior predictive check was used for model evaluation. A linear 2-compartment first-order elimination model with zero-order absorption and lag-time best described the disposition of valsartan. Allometric scaling and standard covariate analysis revealed that age and body size have similar influence; however, after adjustment for body size using fat free mass (FFM), the effect of increasing age was no longer significant on valsartan clearance (2% per year relative to a typical 8 year old with FFM of 30 kg). The population pharmacokinetic model reveals that increase in age has minimal influence on body size dependent clearance of valsartan in children.