Pharmacokinetic and Pharmacodynamic Modelling of the Anti-staphylococcal Activity of AFN-1252 in Humans N. Kaplan1, J. R. Koup2 and B. Hafkin3 Affinium Pharmaceuticals, 1Toronto, ON, Canada and 3Austin, TX, USA; 2The Ann Arbor Pharmacometrics Group, Ann Arbor, USA [email protected]

Introduction and Purpose Results

AFN-1252 is a novel inhibitor of the bacterial fatty acid biosynthesis pathway, specifically targeting Simulation of AFN-1252 Human PK and Anti-Staphylococcal Activity the staphylococcal FabI enzyme. AFN-1252 is a specific spectrum agent with highly potent activity Mouse Net Growth PK/PD Model for Bacterial Growth Response to AFN-1252 against staphylococci and has an MIC90 value of 0.016 μg/ml against all Staphylococcus aureus tested to date (>2,400 strains) including all known -resistant phenotypes. AFN-1252 shows Standard 10 2 Error of the poor or no activity against other bacterial species tested to date. r =0.89 Parameter Estimate Estimate (SEE)

EC50, ng/ml (Total Plasma Conc., ng/mL) 190 41.5 AFN-1252 has demonstrated potent oral in multiple mouse models of infection including 8 95% CI EC50 (Total Plasma Conc., ng/mL) 109 to 271 thigh abscess, skin abscess, pneumonia and peritonitis / septicemia. AFN-1252 is currently in KK, maximum Kill Rate Constant, hr-1 1.5 0.044 clinical development, with a novel oral tablet formulation, as a specific-spectrum antibiotic for KK, half-life hr. 0.462 susceptible and drug-resistant staphylococcal infections. 6 Gamma, Hill Slope 2.07 0.252 KG, Maximum Net Growth Rate Constant, hr-1 0.289 0.015 logCFU/thigh Observed KG, half-life, hr. 2.4 The purpose of this study was to predict the anti-staphylococcal pharmacodynamic effects of AFN- Nmax, Log CFU at 1/2 KG 17.6 0.999 4 KG factor for Study 1 (fold of Study 2 KG value) 0.854 0.086 1252 in humans. Pharmacokinetic (PK) and pharmacodynamic (PD) data from two neutropenic 4 6 8 10 Predicted logCFU/thigh Vehicle Additive Residual Error, ± Log-CFU 0.578 mouse thigh abscess infection models were used to develop a bacterial net growth PK/PD model. Treatment Additive Residual Error, ± Log-CFU 0.455 The primary variable between the two mouse models was drug half-life due to different oral • Model parameters and EC were well estimated with low SEE and residual error values formulations. A population PK model was constructed with AFN-1252 oral Phase 1 PK data that 50 was then applied to the net growth PK/PD model to predict the AFN-1252 PD effects in humans. Methods AFN-1252 NONMEM Human Population PK Model From QD, Single Dose, Oral Data AFN-1252 powder (tosylate monohydrate salt form) for dosing as an oral suspension in the mouse • Individual simulated human PK and anti-staphylococcal activity (color by subject) Standard PK/PD studies, and an immediate release tablet (free base) for oral dosing in the human study Error of the • Efficacy modelled with the AFN-1252 S. aureus MIC90 of 0.016 µg/ml Parameter Estimate Estimate were provided by Affinium Pharmaceuticals. 4000 r2=0.97 (SEE) F1power (F1 = Mean F1 * (250/DOSE)^F1power) 0.348 0.0693 Dose Response of Simulated AFN-1252 Human Anti-staphylococcal Activity S. aureus 29213 was the test organism in the two neutropenic mouse thigh abscess models. In 3000 Mean F1 (@ 250 mg) 0.5 Fixed Absorption Lag Time, hr. 0.157 0.0111 4 Simulated efficacy at 96 hours Study 1 AFN-1252 half lives of 6-8 hours were obtained with 80% PEG400 as the oral suspension KApower (KA = Mean KA * (250/DOSE)^KApower) 0.243 0.201 2000 Mean KA (@ 250 mg), Absorption rate constant hr-1 0.85 0.0681 vehicle. In Study 2 AFN-1252 half lives of 1-2 hours were obtained with 1% Poloxamer 407 as the 2 IIV KA1 % CV 61.5 oral suspension vehicle. Mouse PD (CFU per thigh vs. time) and PK data in infected animals from CL/F, drug L/hr 3.62 0.148 both studies were used to estimate bacterial growth parameters and AFN-1252 PD effects with the 1000 IIV for CL % CV 27.1 0 bacterial net growth PK/PD model using a modification of Campion et al. (Antimicrob. Agents V/F, drug , L 47.9 1.56 IIV for VD % CV 20.3 -2

Observed Plasma Concentration (ng/ml) Concentration Plasma Observed 0 Chemother. 49: 209-219, 2005). The Champion Model describes the rate of change of CFU (not 0 1000 2000 3000 4000 Correlation of CL and VD 0.607 γ γ γ

Log-CFU) as follows: dCFU/dt = KG((Nmax – CFU)/Nmax)*CFU –KK*(Cp /( Cp + EC50 )) Individual Predicted Plasma Concentration (ng/ml) Proportional residual error, % CV 10.5 hr logCFU/gram 96 at -4  where KG is the first order net growth rate constant (hr-1), Nmax is the CFU value at which KG Additive residual error, ± ng/mL 66.3 IIV is the abbreviation for inter-individual variation. -6 would be reduced by half, KK is the first order drug killing rate constant (hr-1), Cp is the plasma 0 100 200 300 400 concentration of drug at a given time, EC50 is the drug concentration at which the killing rate • Model parameters were well estimated with low SEE and residual error values Dose (mg) constant will be half of its maximal value (KK), and γ (Gamma) is the drug slope factor. • Efficacy expressed as change logCFU/gram compared to time 0 (106 CFU/gram) • All individual subject data presented as Box-whisker plots Human PK data were from an oral single ascending dose Phase 1 study utilizing an immediate • Efficacy dose response curve fitted with an inhibitory effect Imax model release tablet of AFN-1252. Population human plasma concentrations were modelled from cohorts receiving 100, 200, 300 and 400 mg AFN-1252. The model was then expanded to predict multiple dose PK for 4 days (at steady-state). NONMEM version 6.1 was used employing a linear one- Conclusions compartment disposition model with first order absorption, where both the rate and extent of absorption were dose dependent. • The bacterial net growth PK/PD model successfully integrated two mouse efficacy studies with different AFN-1252 drug elimination half lives

The human population PK model output was then used to predict individual human subject anti- • Combination of modelled human PK and mouse PD allowed for robust prediction of AFN-1252 efficacy in humans staphylococcal activity at the AFN-1252 MIC against S. aureus (0.016 µg/ml) based on the mouse 90 • Significant anti-staphylococcal activity of AFN-1252 in humans was predicted at QD doses ≥200 mg bacterial net growth PK/PD model. ECCMID 2011, Milan, Italy P 1518