The strategies and experimental methods for implementation of process analytical technology (PAT) on the mobile pharmaceutical manufacturing system, Pharmacy on Demand (PoD), are discussed. With multiple processes to be monitored on the PoD end-to-end continuous manufacturing process, PAT and its real-time process monitoring capability play a significant role in ensuring final product quality. Here, we discuss PAT implementation for real-time monitoring of an intermediate and API concentrations with in-line Fourier-transformed infrared and Raman spectroscopy for the five-step continuous synthesis of ciprofloxacin on the PoD synthesis unit. Two partial least squares regression models were built and verified with flow chemistry experiments to obtain a root-mean-square error of prediction (RMSEP) of 2.2 mg/mL with a relative error of 2.8% for the step 2 FlowIR model and a RMSEP of 0.9 mg/mL with a relative error of 2.8% for the step 5 Raman model. These models were deployed during an 11 h step 1–3 and a 5 h step 4–5 continuous ciprofloxacin synthesis run performed on the PoD system. In these runs, the real-time prediction of intermediate and product concentration was achieved with an online model processing software (Solo_Predictor) and a PAT data collection and management software (synTQ).

The experimental approach taken and challenges overcome in developing a high-purity production (>100 g) scale process for the telescoped synthesis of the antibiotic ciprofloxacin is outlined. The process was first optimized for each step sequentially with regard to purity and yield, with necessary process changes identified and implemented before scaling for longer runs. These changes included implementing a continuous liquid–liquid extraction (CLLE) step and eliminating and replacing the base 1,8-diazabicyclo[5.4.0]­undec-7-ene (DBU) initially used in the ring-closure step due to DBU plausibly forming a decomposition side product that negatively impacted the final product purity. Process conditions were scaled 1.5–2-fold in order to enable the ultimate project goal of producing enough crude ciprofloxacin within 24 h to manufacture 1000 250 mg tablets. Working toward this goal, several production-scale runs were carried out to assess the reproducibility and robustness of the finalized process conditions, with the first three steps being run continuously up to 22 h and the last two steps being run continuously up to 10 h. The end result is a process with a throughput of ∼29 g/h (∼700 g/24 h) with a crude product stream profile of 94 ± 2% and 34 ± 3 mg/mL after five chemical transformations across four reactors and one continuous CLLE unit operation with each intermediate step maintaining a purity >95% by HPLC.

No abstract

No abstract