(423c) A General Framework for Real-Time Risk Assessment in Pharmaceutical Processes: Application to the Automatic Rejection of Off-Spec Oral Drug Doses
- Conference: AIChE Annual Meeting
- Year: 2019
- Proceeding: 2019 AIChE Annual Meeting
- Group: Pharmaceutical Discovery, Development and Manufacturing Forum
Tuesday, November 12, 2019 - 4:12pm-4:33pm
general framework for real-time risk assessment in pharmaceutical processes:
application to the automatic rejection of off-spec oral drug doses
Francesco Rossi*, Sudarshan
Ganesh, Linas Mockus, Gintaras Reklaitis
Forney Hall of Chemical Engineering, 480 Stadium Mall Drive, West Lafayette, IN
47907-2100, United States
A risk assessment
framework is a set of methods and tools for quantitative estimation of the risk
associated with adverse events, which have potential to cause damage to people,
property and the environment. Since risk is intrinsic to most human activities,
risk assessment finds application in several different areas, e.g. finance (Iyer
and Purkayastha, 2017), medicine (Leung et al., 2017) and geology (Cruden and
Fell, 2017). In this contribution, we will focus our attention on a specific
area, namely, the pharmaceutical industry.
In the context of
pharmaceutical processes, risk can be defined as the probability of producing
and distributing drug products that do not satisfy the quality requirements,
imposed by the Food and Drug Administration (FDA). Therefore, it is evident
that risk assessment must already play a very important role in the
pharmaceutical sector. Indeed, it is currently used to assess whether critical
quality attributes of the final drug products, produced by pharmaceutical
plants, meet the FDA regulations (Figure 1). We refer to this type
of risk assessment as offline risk assessment, as product quality is monitored
only at the very end of all of the manufacturing stages.
page-break-after:avoid"> margin-bottom:6.0pt;margin-left:0cm;text-align:center">Figure 1: Conventional
use of risk assessment in the pharmaceutical industry.
The use of offline risk
assessment enables us to detect the presence of off-spec drug products but only
after an entire batch of material has been produced, thus may lead to unnecessary
economic losses and delays in meeting demand. To mitigate this problem, we
propose to use risk assessment in real time, so as to detect and automatically
reject off-spec material before it contaminates on-spec production (Figure 2). We refer to this type
of risk assessment as online risk assessment, as product quality is
continuously monitored over time, during all manufacturing stages.
page-break-after:avoid"> margin-bottom:6.0pt;margin-left:0cm;text-align:center">Figure 2: Innovative
use of risk assessment in pharmaceutical processes.
The systematic use of online
risk assessment may benefit both continuous and batch drug manufacturing plants
by decreasing their environmental impact, lowering drug production costs and reducing
the probability of delays in production (disposal of off-spec drug products
often requires incineration, which is expensive and has non-negligible carbon
footprint). On the other hand, frameworks for real-time risk assessment require
efficient computational methods for estimation of probability distributions
(PDFs) and for uncertainty propagation through systems of nonlinear equations,
which may not be readily available. In this contribution, we overcome this difficulty
by utilizing the new, efficient PDF estimation strategies, proposed by Rossi,
et al. (2018).
The potential of
real-time risk assessment is demonstrated on a pilot-scale drug product
manufacturing plant for production of oral drug doses, equipped with an OSIsoft
PI system that enables us to retrieve data from equipment sensors, in-line and
at-line sensors. The results of our simulations show that the systematic use of
online risk assessment can improve process economics and reduce the amount of
waste, produced by the aforementioned pilot-scale plant. We also demonstrate
that real-time risk assessment is superior to other conventional techniques for
online detection and rejection of off-spec products, which rely on dynamic data
reconciliation and/or residence time distribution.
Cruden, D.M., Fell, R.
(2017). Landslide risk assessment. Taylor & Francis, New York (NY).
Iyer, K.C., Purkayastha,
D. (2017). Credit risk assessment in infrastructure project finance: relevance
of credit ratings. Journal of Structured Finance, 22, 17-25.
Daskalopoulou, S.S., Dasgupta, K., McBrien, K., Butalia, S., Zarnke, K.B., Gelfer,
M. (2017). Hypertension Canada's 2017 guidelines for diagnosis, risk
assessment, prevention, and treatment of hypertension in adults. Canadian
Journal of Cardiology, 33, 557-576.
Rossi, F., Mockus, L.,
Manenti, F., Reklaitis, G. (2018). Assessment of accuracy and computational
efficiency of different strategies for estimation of probability distributions
applied to ODE/DAE systems. Computer Aided Chemical Engineering, 44, 1543-1548.