Quality by design approach for early understanding of active pharmaceutical ingredient recovery process through dead-end filtration

E. Ojo, ACME. Rayat, C. Brown, C. Price, M. Hoare, A. Florence, Andrea Johnston

Research output: Contribution to conferencePosterpeer-review

Abstract

This study applied some concepts of quality-by-design (QbD) approach for an early understanding of crystal recovery through filtration. A bespoke laboratory scale dead-end filtration platform (modified Biotage vacuum master (BVM) ≈ 50 mL working volume) was used to investigate the recovery of an active pharmaceutical ingredient (API) of different size distributions using acetaminophen crystals (micronised, medium-sized Bioxtra and coarse) as a case study. The method involved: (1) identification of critical process parameters (CPPs) with significant impact on process stability (a process risk evaluation step based on one factor at a time); (2) design of experiment to screen the influence of design factors (such as filter pore size, pressure difference, crystal loading and particle size distribution (PSD)) contributing to process instability based on process responses (volumetric flux and specific cake resistance); and (3) investigate the optimal process window for reduced probability of failure and process predictability. The filtration process responses were characterised by assessing the filtrate flow rate by the application of Darcy’s law. Initial assessments of process steadiness for acetaminophen crystals recovery shows that the filter pore size and API crystal sizes are critical. A non-linear process dependency was observed between the applied pressure difference, pore size, and crystal size. Screening crystal recovery conditions based on the design of experiment (DoE) approach indicated a steady filtration process for all crystal sizes tested except for coarse crystals which shows non-valid (negative) specific cake resistance at conditions of 5 µm pore size filter and 100 mbar pressure difference. However, pressure difference and pore size had a notable impact on the process responses. As a result, 10 µm pore size filter was used for investigating the optimal process window. Process predictability was demonstrated by data clustering and refinement based on partial least square model. The results showed good predictability with >98% regression between the predicted and experimental data. Verification of optimal operating window with less than 5% probability failure resulted in conditions of 300 – 450 mbar pressure difference and PSD of 45 – 110 µm. The approach studied using the small-scale BVM provided an early data gathering and systematic approach to understanding process interactions affecting crystal recovery through dead-end filtration.
Original languageEnglish
Number of pages1
Publication statusPublished - 9 Nov 2019
EventThe American Association of Pharmaceutical Scientists (AAPS) 2019 PharmaSci 360 - Henry B. Gonzalez Convention Center, San Antonio, TX., United States
Duration: 3 Nov 20196 Nov 2019

Conference

ConferenceThe American Association of Pharmaceutical Scientists (AAPS) 2019 PharmaSci 360
Abbreviated titleAAPS
Country/TerritoryUnited States
CitySan Antonio, TX.
Period3/11/196/11/19

Cite this