DeepARV: ensemble deep learning to predict drug-drug interaction of clinical relevance with antiretroviral therapy

Drug-drug interaction (DDI) may result in clinical toxicity or treatment failure of antiretroviral therapy (ARV) or comedications. Despite the high number of possible drug combinations, only a limited number of clinical DDI studies are conducted. Computational prediction of DDIs could provide key evidence for the rational management of complex therapies. Our study aimed to assess the potential of deep learning approaches to predict DDIs of clinical relevance between ARVs and comedications. DDI severity grading between 30,142 drug pairs was extracted from the Liverpool HIV Drug Interaction database. Two feature construction techniques were employed: 1) drug similarity profiles by comparing Morgan fingerprints, and 2) embeddings from SMILES of each drug via ChemBERTa, a transformer-based model. We developed DeepARV-Sim and DeepARV-ChemBERTa to predict four categories of DDI: i) Red: drugs should not be co-administered, ii) Amber: interaction of potential clinical relevance manageable by monitoring/dose adjustment, iii) Yellow: interaction of weak relevance and iv) Green: no expected interaction. The imbalance in the distribution of DDI severity grades was addressed by undersampling and applying ensemble learning. DeepARV-Sim and DeepARV-ChemBERTa predicted clinically relevant DDI between ARVs and comedications with a weighted mean balanced accuracy of 0.729 ± 0.012 and 0.776 ± 0.011, respectively. DeepARV-Sim and DeepARV-ChemBERTa have the potential to leverage molecular structures associated with DDI risks and reduce DDI class imbalance, effectively increasing the predictive ability on clinically relevant DDIs. This approach could be developed for identifying high-risk pairing of drugs, enhancing the screening process, and targeting DDIs to study in clinical drug development.

Physiologically-based pharmacokinetic modelling of long-acting injectable cabotegravir and rilpivirine in pregnancy

AIMS

Long-acting cabotegravir and rilpivirine have been approved to manage HIV in adults, but data regarding safe use in pregnancy are limited. Physiologically-based pharmacokinetic (PBPK) modelling was used to simulate the approved dosing regimens in pregnancy and explore if Ctrough was maintained above cabotegravir and rilpivirine target concentrations (664 and 50 ng/mL, respectively).

METHODS

An adult PBPK model was validated using clinical data of cabotegravir and rilpivirine in nonpregnant adults. This was modified by incorporating pregnancy-induced metabolic and physiological changes. The pregnancy PBPK model was validated with data on oral rilpivirine and raltegravir (UGT1A1 probe substrate) in pregnancy. Twelve weeks' disposition of monthly and bimonthly dosing of long-acting cabotegravir and rilpivirine was simulated at different trimesters and foetal exposure was also estimated.

RESULTS

Predicted Ctrough at week 12 for monthly long-acting cabotegravir was above 664 ng/mL throughout pregnancy, but below the target in 0.5% of the pregnant population in the third trimester with bimonthly long-acting cabotegravir. Predicted Ctrough at week 12 for monthly and bimonthly long-acting rilpivirine was below 50 ng/mL in at least 40% and over 90% of the pregnant population, respectively, throughout pregnancy. Predicted medians (range) of cord-to-maternal blood ratios were 1.71 (range, 1.55-1.79) for cabotegravir and 0.88 (0.78-0.93) for rilpivirine between weeks 38 and 40.

CONCLUSIONS

Model predictions suggest that monthly long-acting cabotegravir could maintain antiviral efficacy throughout pregnancy, but that bimonthly administration may require careful clinical evaluation. Both monthly and bimonthly long-acting rilpivirine may not adequately maintain antiviral efficacy in pregnancy.

General Framework to Quantitatively Predict Pharmacokinetic Induction Drug-Drug Interactions Using In Vitro Data

INTRODUCTION

Metabolic inducers can expose people with polypharmacy to adverse health outcomes. A limited fraction of potential drug-drug interactions (DDIs) have been or can ethically be studied in clinical trials, leaving the vast majority unexplored. In the present study, an algorithm has been developed to predict the induction DDI magnitude, integrating data related to drug-metabolising enzymes.

METHODS

The area under the curve ratio (AUC_ratio) resulting from the DDI with a victim drug in the presence and absence of an inducer (rifampicin, rifabutin, efavirenz, or carbamazepine) was predicted from various in vitro parameters and then correlated with the clinical AUC_ratio (N = 319). In vitro data including fraction unbound in plasma, substrate specificity and induction potential for cytochrome P450s, phase II enzymes and uptake, and efflux transporters were integrated. To represent the interaction potential, the in vitro metabolic metric (IVMM) was generated by combining the fraction of substrate metabolised by each hepatic enzyme of interest with the corresponding in vitro fold increase in enzyme activity (E) value for the inducer.

RESULTS

Two independent variables were deemed significant and included in the algorithm: IVMM and fraction unbound in plasma. The observed and predicted magnitudes of the DDIs were categorised accordingly: no induction, mild, moderate, and strong induction. DDIs were assumed to be well classified if the predictions were in the same category as the observations, or if the ratio between these two was < 1.5-fold. This algorithm correctly classified 70.5% of the DDIs.

CONCLUSION

This research presents a rapid screening tool to identify the magnitude of potential DDIs utilising in vitro data which can be highly advantageous in early drug development.

Evaluation of drug-drug interaction between rilpivirine and rifapentine using PBPK modelling

Tuberculosis remains the leading cause of death among people living with HIV. Rifapentine is increasingly used to treat active disease or prevent reactivation, in both cases given either as weekly or daily therapy. However, rifapentine is an inducer of CYP3A4, potentially interacting with antiretrovirals like rilpivirine. This in silico study investigates the drug-drug interaction (DDI) magnitude between daily oral rilpivirine 25 mg with either daily 600 mg or weekly 900 mg rifapentine. A physiologically based pharmacokinetic (PBPK) model was built in Simbiology (Matlab R2018a) to simulate the drug-drug interaction. The simulated PK parameters from the PBPK model were verified against reported clinical data for rilpivirine and rifapentine separately, daily rifapentine with midazolam, and weekly rifapentine with doravirine. The simulations of concomitant administration of rifapentine with rilpivirine at steady-state lead to a maximum decrease on AUC_0-24 and C_trough by 83% and 92% on day 5 for the daily rifapentine regimen and 68% and 92% for the weekly regimen on day 3. In the weekly regimen, prior to the following dose, AUC_0-24 and C_trough were still reduced by 47% and 53%. In both simulations, the induction effect ceased 2 weeks after the interruption of rifapentine's treatment. A daily double dose of rilpivirine after initiating rifapentine 900 mg weekly was simulated but failed to compensate the drug-drug interaction. The drug-drug interaction model suggested a significant decrease on rilpivirine exposure which is unlikely to be corrected by dose increment, thus coadministration should be avoided.

PBPK Modelling of Dexamethasone in Patients With COVID-19 and Liver Disease

The aim of the study was to apply Physiologically-Based Pharmacokinetic (PBPK) modelling to predict the effect of liver disease (LD) on the pharmacokinetics (PK) of dexamethasone (DEX) in the treatment of COVID-19. A whole-body PBPK model was created to simulate 100 adult individuals aged 18–60 years. Physiological changes (e.g., plasma protein concentration, liver size, CP450 expression, hepatic blood flow) and portal vein shunt were incorporated into the LD model. The changes were implemented by using the Child-Pugh (CP) classification system. DEX was qualified using clinical data in healthy adults for both oral (PO) and intravenous (IV) administrations and similarly propranolol (PRO) and midazolam (MDZ) were qualified with PO and IV clinical data in healthy and LD adults. The qualified model was subsequently used to simulate a 6 mg PO and 20 mg IV dose of DEX in patients with varying degrees of LD, with and without shunting. The PBPK model was successfully qualified across DEX, MDZ and PRO. In contrast to healthy adults, the simulated systemic clearance of DEX decreased (35%–60%) and the plasma concentrations increased (170%–400%) in patients with LD. Moreover, at higher doses of DEX, the AUC ratio between healthy/LD individuals remained comparable to lower doses. The exposure of DEX in different stages of LD was predicted through PBPK modelling, providing a rational framework to predict PK in complex clinical scenarios related to COVID-19. Model simulations suggest dose adjustments of DEX in LD patients are not necessary considering the low dose administered in the COVID-19 protocol.

Drug-Drug Interactions in People Living with HIV at Risk of Hepatic and Renal Impairment: Current Status and Future Perspectives

Despite the advancement of antiretroviral therapy (ART) for the treatment of human immunodeficiency virus (HIV), drug–drug interactions (DDIs) remain a relevant clinical issue for people living with HIV receiving ART. Antiretroviral (ARV) drugs can be victims and perpetrators of DDIs, and a detailed investigation during drug discovery and development is required to determine whether dose adjustments are necessary or coadministrations are contraindicated. Maintaining therapeutic ARV plasma concentrations is essential for successful ART, and changes resulting from potential DDIs could lead to toxicity, treatment failure, or the emergence of ARV‐resistant HIV. The challenges surrounding DDI management are complex in special populations of people living with HIV, and often lack evidence‐based guidance as a result of their underrepresentation in clinical investigations. Specifically, the prevalence of hepatic and renal impairment in people living with HIV are between five and 10 times greater than in people who are HIV‐negative, with each condition constituting approximately 15% of non‐AIDS‐related mortality. Therapeutic strategies tend to revolve around the treatment of risk factors that lead to hepatic and renal impairment, such as hepatitis C, hepatitis B, hypertension, hyperlipidemia, and diabetes. These strategies result in a diverse range of potential DDIs with ART. The purpose of this review was 2‐fold. First, to summarize current pharmacokinetic DDIs and their mechanisms between ARVs and co‐medications used for the prevention and treatment of hepatic and renal impairment in people living with HIV. Second, to identify existing knowledge gaps surrounding DDIs related to these special populations and suggest areas and techniques to focus upon in future research efforts.