(586r) Simulation and Global Sensitivity Analysis of Pharmaceutical Processes for Solid Drug Manufacture
- Conference: AIChE Annual Meeting
- Year: 2013
- Proceeding: 2013 AIChE Annual Meeting
- Group: Pharmaceutical Discovery, Development and Manufacturing Forum
- Time: Wednesday, November 6, 2013 - 6:00pm-8:00pm
There are significant economic incentives to transition from batch to continuous processing of pharmaceutical products (1). However there are several challenges to be overcome in order to implement continuous manufacturing systems for solids-based processes. Among these is the need to develop increased understanding of continuous processes and their sensitivity to variability in input materials, processing equipment and operating conditions. The creation of predictive process models can play an important role in developing the requisite process understanding. Process models can be used to reduce experimental requirements, enhance process understanding, and develop control strategies. Modeling is also an important component of the Quality by Design (QbD) paradigm described by the ICH Q8 guidance for pharmaceutical development. (2)
In recent work a computationally efficient approach for the simulation of integrated pharmaceutical processes based on the use of an equation-oriented process simulator (gProms®) has been presented. Within this framework both mechanistic and semi-empirical models for particulate processes havebeen implemented. (3) In the current work this framework is used to model a process for the continuous production of pharmaceutical tablets via direct compaction. Direct compaction is the most straightforward approach to tablet production and is thus an appealing manufacturing route if it can be successfully implemented for the product of interest. Because the direct compaction process does not include a granulation step, the physical properties of the raw materials have the potential to significantly impact performance during compaction. Content uniformity may also be quite sensitive to powder properties and the performance of unit operations such as blending. Therefore it is important to understand the impact of material properties and unit operation design and operating parameters on process performance if direct compaction is to be successfully implemented. To this end sensitivity analysis can be conducted with the dual objectives of model improvement and process understanding.
Sensitivity analysis is an important component of modeling that generally occurs in tandem with process development. Sensitivity analysis can be used to identify critical process parameters and to examine the validity of assumptions underlying the developed process models. In the current work, global sensitivity analysis is used to identify gaps in the existing process model with the goal of defining specific areas where model improvement can be effected. In addition, global sensitivity analysis of the integrated flowsheet simulation is used to assess the impact of variability in input material properties and process design and operating parameters on product quality.
In the current work a global sensitivity analysis is conducted which considers the effect of 22 uncertain design, operating and raw material properties parameters on 19 responses including indicators of unit operation performance and of product quality. A variety of global sensitivity analysis techniques are implemented, including regression based and variance based methods. The partial rank correlation coefficient, a regression based sensitivity metric calculated from rank transformed data, has been used for preliminary sensitivity analysis. PRCCs are most accurate as an indicator of sensitivity when the inputs are not highly correlated and the responses vary monotonically with the inputs. (4) In this case the majority of input-response relationships are monotonic or nearly monotonic, but some of the model inputs are correlated. Thus PRCCs alone may not be sufficient to guarantee accurate assessment of sensitivities. In addition, the partial rank correlation coefficient does not provide an indication of the extent to which interactions among parameters may influence the process. This can be addressed using the extended FAST (eFAST) method, which gives variance based sensitivity and total sensitivity indices. (5), (6) These sensitivity metrics are valid for input-output relationships that range from linear to nonlinear and non-monotonic and are thus more widely applicable than PRCCs. In addition, the total sensitivity indices indicate the sensitivity of a model output to all higher order interactions of a process parameter with all other inputs. However, eFAST does not provide information about specific interacting inputs which give rise to the total sensitivity indices. For this, random sampling HDMR (RS-HDMR) can be used. RS-HDMR yields first order sensitivity indices and second order indices for specific parameter interactions. (7), (8) (9) In this case RS-HDMR is used as a supplement to the extended FAST method in order to determine if there are any specific parameter interactions that contribute significantly to total sensitivity for a given output.
The results of global sensitivity analysis have been used to suggest potential model improvements and to identify process parameters that may be deemed critical to ensuring product quality. For instance, the lubricant concentration does not significantly affect the simulated tablet hardness. However it is known in practice that both lubricant and API concentration can result in varying tablet properties. Therefore a model modification to account for the impact of tablet composition on hardness is suggested to improve the predictive ability of the tablet compaction model. Tablet weight and composition are found to be highly sensitive to the API and excipient feeder parameters and raw material bulk densities. Thus sensitivity analysis can also contribute to determination of critical process parameters. These variables should be manipulated in order to successfully implement a control strategy for tablet API concentration.
1. Economic Analysis of Integrated Continuous and Batch Pharmaceutical Manufacturing: A Case Study. Schaber, S.D., Gerogiorgis, D.I., Ramachandran, R., Evans, J.M.B., Barton, P.I., Trout, B.L.17, 2011, Ind. Eng. Chem. Res., Vol. 50, pp. 10083–10092.
2. ICH Harmonised Tripartite GuidelinePHARMACEUTICAL DEVELOPMENT Q8(R2) Current Step 4 version. s.l. : International Conference on Harmonisation of Technical Requirements for Registration of pharmaceuticals for human use, August 2009.
3. Reduced-order discrete element method modeling. Boukouvala, F., Gao, Y., Muzzio, F., Ierapetritou, M.G.Chemical Engineering Science, p. Article in Press.
4. Saltelli, A.K., Chan, K., Scott, M. Sensitivity Analysis. s.l. : John Wiley & Sons Inc., 200.
5. An alternative way to compute Fourier Amplitude Sensitivity Test (FAST). 1998, Comput. Statist. Data Anal., Vol. 26, pp. 445-460.
6. A quantitative, model independent method for global sensitivity analysis of model output. Saltelli, A., Tarantola, S., Chan, K.1999, Technometrics, Vol. 41, pp. 39-56.
7. Regularized random-sampling high dimensional model representation (RS-HDMR). Li, G., Rabitz, H.3, 2008, Journal of Mathematical Chemistry, Vol. 43, p. 1207.
8. Global uncertainty assessments by high dimensional model representations (HDMR). Li, G., Wang, S.W., Rabitz, H., Wang, S., Jaffé,P.2002, Chemical Engineering Science, Vol. 57, pp. 4445 – 4460.
9. GUI–HDMR – A software tool for global sensitivity analysis of complex models. Ziehn, T., Tomlin, A.S.2009, Environmental Modelling & Software, Vol. 24, pp. 775-785.
10. Computer-Aided Flowsheet Simulation of a Pharmaceutical Tablet Manufacturing Process Incorporating Wet Granulation. Boukouvala, F., Chaudhury, A., Sen, M., Zhou, R., Mioduszewski, L., Ierapetritou, M.G., Ramachandran, R.2013, J Pharm Innov, Vol. 8, pp. 11-27.
11. An integrated approach for dynamic flowsheet modeling and sensitivity analysis of a continuous tablet manufacturing process. Boukouvala, F., Niotis,V., Ramachandran, R. Muzzio, F.J., Ierapetritou, M.G.2012, Computers and Chemical Engineering, Vol. 42, pp. 30-47.
12. Sensitivity analysis for nonlinear mathematical models. Sobol, I.M.1993, Mathematical Modelling & Computational Experiment, Vol. 1, pp. 407-414.
13. A Global Sensitivity Study of Sulfur Chemistry in a Premixed Methane Flame Model Using HDMR. Ziehn, T., Tomlin, A.S.s.l. : Wiley Periodicals, Inc., 2008, Int J Chem Kinet, Vol. 40, pp. 742–753.
14. A Quantitative Model-Independent Method for Global Sensitivity Analysis of Model Output. Saltelli, A., Tarantola, S., Chan, K.P.-S. 1, 1999, Technometrics, Vol. 41, pp. 39-56.
This paper has an Extended Abstract file available; you must purchase the conference proceedings to access it.
Do you already own this?
Log In for instructions on accessing this content.
2013 AIChE Annual Meeting
|AIChE Graduate Student Members||Free|
|AIChE Undergraduate Student Members||Free|
Pharmaceutical Discovery, Development and Manufacturing Forum only
|AIChE Food, Pharmaceutical & Bioengineering Division Members||Free|
|AIChE Graduate Student Members||Free|
|AIChE Undergraduate Student Members||Free|