Identification of long-acting NRTI candidates through in silico

modelling

Rajith KR Rajoli1, Steve Rannard2, Charles Flexner3, Andrew Owen1, Marco Siccardi1

1 Molecular and Clinical Pharmacology, University of Liverpool, Liverpool, 2 Department of Chemistry, University of Liverpool, Liverpool, 3 Johns Hopkins University, Baltimore, Maryland

Introduction

• Currently only a limited number of LA drugs exist

• The development of complimentary LA formulations of multiple ARV classes will expand regimens to better manage therapy and prevention

• The identification of suitable LA candidates is a complex process and multiple experimental and clinical studies are essential

• Computer based simulations can represent a valuable tool to predict ARV pharmacokinetics and pharmacodynamics

• to identify potential NRTI candidates for LA strategies through a rational evaluation of pharmacokinetics and pharmacodynamics

• integration of pharmacodynamic data (NIAID ChemDB) and computational pharmacokinetics modelling to simulate PK/PD of candidate LA agents

Aims

Pharmacodynamic data

Division of AIDS Anti-HIV/OI/TB Therapeutics Database - https://chemdb.niaid.nih.gov/

Prediction of key pharmacokinetics parameters through QSAR

• Quantitative Structure-Activity Relationship (QSAR) is a mathematical relationship between biological activity of a molecule and its physicochemical and geometrical descriptors and properties

• The relationship between biological activity and molecular properties could be used to evaluate new compounds

Molecular descriptors

Descriptor Definition

log P Octanol-water partition coefficient

pKa Dissociation constant

HBD Number of hydrogen bond donors

PSA Polar surface area

SaasC Sum of (aasC–) electro-topological states

SssCH2_acnt Count of (– CH –) electro-topological states

SaasC_acnt Count of (aasC–) electro-topological states

SdssC_acnt Count of (= C < ) electro-topological states

Gmax Maximum E-state value of an atom in a molecule

Hmin Maximum hydrogen E-state value of an atom in a molecule

Hmax Maximum hydrogen E-state value of an atom in a molecule

MaxNeg The largest negative charge over the atoms in a molecule

MaxQP The largest positive charge over the atoms in a molecule

log PpKa

HBD

PSA

SaasC

SaasC_acnt

Gmax

Hmin

MaxNeg

Hmax

Renal Clearance1

Blood-to-plasma ratio2

Volume of distribution3

1Dave RA, Morris ME. Drug Metabolism and Disposition. 2015;43(1):73-81.2Paixão P, Gouveia LF, Morais JAG. European Journal of Pharmaceutical Sciences. 2009 3/2/;36(4–5):544-543Poulin et al. Journal of Pharmaceutical Sciences. 2002;91:129-156

1. Blood to plasma ratio

2. Fraction unbound

3. Plasma stability

4. Renal clearance

5. Volume of distribution

QSAR models for

Integration of QSAR predictions in PBPK

NIAID ChemDBscreening

QSAR models for the prediction of key

variables

PHARMACOKINETICS PHARMACODYNAMICS

Identification of NRTI candidates

Computational environment to

generate molecular predictors

Application of QSAR/PBPK models

to predict theoretical PK

Qualification of preliminary simulations

(characterised ARVs)

Simulation of theoretical LA

strategies

Project strategy

Results - Modelling qualification for existing NRTIs

Renal clearance (L/hr)1

Observed Calculated Ratio (c/o)

Tenofovir 11.342 19.82 1.75

Emtricitabine 12.803 27.42 2.14

Acyclovir 14.994 25.16 1.68

Zidovudine 20.815 30.67 1.47

Stavudine 16.326 27.69 1.70

Lamivudine 11.987 27.32 2.28

Volume of distribution (L/kg)

Observed Calculated Ratio (c/o)

Tenofovir 0.812 0.79 0.98

Emtricitabine 0.803 0.57 0.71

Acyclovir 0.6810 0.52 0.76

Zidovudine 1.605 0.48 0.30

Stavudine 0.606 0.58 0.97

Lamivudine 1.307 0.47 0.36

Renal clearance

1. Dave RA, Morris ME. Drug Metabolism and Disposition 2015; 43: 73-81.2. DrugBank. Tenofovir. http://www.drugbank.ca/drugs/DB00300 3. DrugBank. Emtricitabine. http://www.drugbank.ca/drugs/DB00879 4. Laskin OL, de Miranda P, King DH et al. Antimicrobial Agents and Chemotherapy 1982; 21: 804-7.5. DrugBank. Zidovudine. https://www.drugbank.ca/drugs/DB004956. DrugBank. Stavudine. https://www.drugbank.ca/drugs/DB00649 7. DrugBank. Lamivudine. https://www.drugbank.ca/drugs/DB00709

Results – simulation of quarterly LA administration

343654 343656

0

10

20

30

40

50

60

70

80

90

0 14 28 42 56 70 84

Co

nce

ntr

atio

n (

ng/

ml)

Time (days)

Mean Mean ± SD

EC50 – 0.03 ng/ml

Dose – 2000 mg Release rate – 5 x 10-4 h-1

0

10

20

30

40

50

60

70

80

90

0 14 28 42 56 70 84

Co

nce

ntr

atio

n (

ng/

ml)

Time (days)

Mean Mean ± SD

EC50 – 0.22 ng/ml

Results

Shortlisted compounds including therapeutic index, EC50, simulated Ctrough and relative ratio for monthly and quarterly LA injections

Monthly Quarterly

AIDS # Therapeutic index EC50 (ng/ml) Ctrough (ng/ml) Ratio (Ctrough/EC50) Ctrough (ng/ml) Ratio (Ctrough/EC50)

343654 1000000 0.03 73.2 2495.3 25.6 873.7

105173 43667 0.19 65.4 341 22.8 119.1

343656 3000000 0.22 77.9 359.5 24.6 113.5

168640 48500 0.37 32.3 88.2 10.2 27.8

108530 44000 0.35 28.1 81.5 8.9 25.8

113361 50417 0.99 50.1 50.5 15.8 16

212706 42571 1.6 69.7 44.4 22 14

Results

AIDS # Class Notes Chemical name

343654 Purine Nucleosides EFdA 4'-ethynyl-2-fluoro-2'-deoxyadenosine

105173 Pyrimidine Nucleosides 3'-Azido-3'-deoxythymidin-5'-yl O-(4-hydroxybutyl) carbonate

343656 Purine Nucleosides ECldA 9H-Purin-6-amine

168640 Triazines4-[[4-amino-6-[(2-chloro-4,6-dimethylphenyl)amino]-1,3,5-triazin-2-

yl]amino]benzonitrile

108530 Triazines4-({3-Amino-5-[(2,4,6-trimethylphenyl)amino]1,3,5-triazin-2-

yl}amino)benzenecarbonitrile

113361 Pyrimidine Nucleotides Phosphonate deriv. of AZT 5'-cyclohexyl phosphonate

212706 Pyrimidine Nucleosides

Carbonic acid (2S,3S,5R)-3-azido-5-(5-methyl-2,4-dioxo-3,4-dihydro-2H-pyrimidin-1-yl)-tetrahydro-furan-2-ylmethyl ester (2R,3R,5S)-3-azido-5-(5-

methyl-2,4-dioxo-3,4-dihydro-2H-pyrimidin-1-yl)-tetrahydro-furan-2-ylmethyl ester

Limitations

• Quantitative Structure-Activity Relationship (QSAR) modelling is characterised by some intrinsic limitations

• Lymphatic circulation and activity of efflux and influx transporters at the site of injection have not been considered.

• The technological complexities associated with reformulation can represent a challenging barrier

• QSAR/PBPK predictions will require confirmatory experimental studies for a more precise prediction of pharmacokinetics

Summary

• A novel QSAR/PBPK model for NRTI was developed and qualified against

existing clinical data

• Integrated PBPK/QSAR models assisted in identifying potential NRTI

candidates for LA delivery

• The QSAR/PBPK approach can be applied to prioritise further research for

potential candidates

• This rational approach for the selection of suitable candidates may prove

useful to support the development of additional LA formulations across

multiple ARV classes.

Andrew OwenSteve RannardDavid BackRajith Kumar ReddyNeill LiptrottTom McDonaldSaye KhooOwain RobertsLaura DIckinsonLee TathamJames HobsonPaul CurleyAdeniyi OlagunjuMegan NearyChristina ChanJustin ChiongLaura ElseHenry PertinezAlessandro SchipaniMarco GiardielloFiona HattonSam AutyAndy Dwyer

Adny Henrique Silva

Kim ScarsiSusan SwindellsAnthony Podany

Jose Molto

Catia MarzoliniManuel Battegay

Marta Boffito

Charles FlexnerCaren Meyers

Acknowledgments