Regulatory Aspects of a cGMP Continuous Drug Substance Process and Contrasts with Continuous Drug Product

Regulatory Aspects of a cGMP Continuous Drug Substance Process and Contrasts with Continuous Drug Product
Continuous drug substance processes were run in a new building that won the International Society for Pharmaceutical Engineering 2019 Facility of the Year Award for Innovation. The talk focuses on the benefits of continuous drug substance processing, regulatory aspects, and contrasts with continuous drug product. There was no startup transition waste 10 out of 13 unit operations. Three different continuous reactions all started with a planned amount of startup transition waste, but product collection started before steady state for all of them. The continuous extractions, evaporations, crystallization, semi-continuous filtration, and one of the reactions all started in a controlled procedural manner without diverting any material to waste. Processes were designed for ease of stop/restart with no diverting to waste. Run time in the plant (>1 month) was longer than in development. Surge tanks decoupled upstream and downstream continuous sections, and also facilitated the use of rolling average approach for quantifying acceptable magnitude and duration of disturbances. PAT frequency was less than DP CM, enabled because of long mean residence times and broad RTDs. High specificity of on-line HPLC was more important than high frequency of spectroscopic probes. PAT was used for the continuous reactions, but only parametric control was used for workup and isolation, including continuous crystallization. Continuous crystallization was needed for kinetic rejection of a key quinoline dimer impurity which could not be rejected thermodynamically. PAT and parametric control both informed divert and return from divert. Batch size ranged from 4 kg to 18 kg API, taking advantage of the flexible batch size benefits of CM. RTD Modeling quantified lot genealogy. Cleaning frequency was adjustable. It was left to the judgment of the cGMP operations and technical staff to decide when it was time to stop for a cleanout, rather than specifying a fixed time duration. It is anticipated that each subsequent run will get longer than the previous before cleanout because of the gained understanding of fouling mitigation over time. Parametric control was used for several continuous processing sections. Development scale was 200X-1000X smaller than manufacturing scale.