Development of a USP-4 IVIVC methodology for LAI antivirals: Case studies with cabotegravir and rilpivirine

18

Development of a USP-4 IVIVC methodology for LAI antivirals: Case studies with cabotegravir and rilpivirine

Emad Anaam, Henry Pertinez, Lee Tatham, Paul Curley, Paul Valentijn, Eduardo Gallardo Toledo, Joanne Sharp and Andrew Owen

University of Liverpool

Background: Long-acting injectables (LAI) offer a promising approach to improving adherence and efficacy in the treatment of HIV. Establishing a reliable in vitro-in vivo correlation (IVIVC) is critical for predicting the in vivo performance of these formulations. This study aimed to develop a USP apparatus 4 (USP-4) IVIVC methodology specifically for LAI antivirals using cabotegravir (CAB) and rilpivirine (RPV) as model compounds.

Material and methods: CAB and RPV formulations were subjected to in vitro release testing using the USP-4 apparatus with dose release from a float-a-lyzer dialysis device. Method optimization investigated adjustment of parameters such as temperature, dialysis membrane MWCO and flow rate, Tween 20 excipient concentration and pH of the circulating buffer. Cumulative release profiles were sampled and analysed over a period of 30 days and fitted with a bi-exponential mathematical model to allow extrapolation to estimate longer release durations. For in vivo analysis, clinical LAI pharmacokinetic (PK) data for CAB and RPV were obtained from literature, and in vivo release profiles derived through deconvolution using intravenous (IV) bolus disposition impulse responses estimated for both drugs via PBPK scaling due to lack of clinical IV data. IVIVC predictions and comparisons were made both by convolution of the in vitro release with IV disposition for direct comparison with the in vivo PK exposure profile and by Levy-plot correlations of in vitro and in vivo cumulative release.

Results: Results obtained show sustained in vitro cumulative release for CAB and RPV over 30 days in the USP-4 system. However, in vitro cumulative release extrapolated for the duration of in vivo PK profiles was lower than cumulative release derived from in vivo data. Concordantly, convolution of the in vitro release profiles with estimated IV bolus disposition response underpredicted observed LAI PK exposure profiles. However, inclusion of a linear multiple scaling factor revealed the shape of the in vivo PK exposure profile was predicted by the in vitro release profile. This is illustrated in the Levy correlation plots showing partial correlation between in vitro and in vivo release but with deviation away from the line of unity.

Conclusions: Initial results from developing a USP-4 methodology for application to LAI formulations and IVIVC to clinical LAI PK exposures indicate potential for prediction of in vivo PK exposure or utility as a means of ranking/comparing formulations by in vitro release. Work is ongoing to further investigate optimization of experimental parameters and to expand the library of LAI formulations being investigated to identify trends across a range of LAI products and confirm generalizability of the methodology.