RESEARCH ARTICLE Development of a Fluid Bed Granulation Design Space Using Critical Quality Attribute Weighted Tolerance Intervals BRIAN M. ZACOUR, JAMES K. DRENNEN III, CARL A. ANDERSON Duquesne Center for Pharmaceutical Technology, Duquesne University, Pittsburgh, Pennsylvania 15282 Received 28 February 2012; revised 27 March 2012; accepted 19 April 2012 Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23185 ABSTRACT: The fluid bed granulation and drying unit operation were used as a case study for control systems implementation. This single processor was used to blend, granulate, dry, and cool the materials. The current study demonstrated control of each of the phases using a fully automated, hybrid control system that incorporated first-principle modeling, empirical design of experiments (DOE), and process analytical technology to assure the production of constant product quality. The system allowed data to be collected efficiently for the develop- ment of a rigorous design space that combined formulation factors, process factors, and their interactions to define a tolerance surface where risk of future product failure was significantly reduced. The DOE incorporated microcrystalline cellulose and lactose monohydrate, excipients with substantially different wetting properties, to elucidate the relationship between the critical process parameters of the unit operation and the material properties of the formulation com- ponents. The extended analysis of covariance model enabled these factors and their interaction terms to be described in a single model. The results indicate that the development of a tolerance interval-based weighted design space can enhance product understanding and thereby help to assure future product quality. © 2012 Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci Keywords: fluid bed processing; process analytical technology; multivariate modeling; design space; quality by design INTRODUCTION The pharmaceutical industry has invested a substan- tial amount of resources in recent years to develop manufacturing systems that offer improved product quality while limiting costs. The United States Food and Drug Administration (FDA) has accepted the guidelines put forth by the International Conference on Harmonization [ICH-Q8(R2)] 1 that allow for op- erational flexibility within a validated design space to allow fully automated systems that incorporate real-time data management to assure finished prod- uct quality. These systems offer continuous improve- ment of the process and resulting drug product by allowing information gained during manufacturing through online process measurements to inform pro- cess adjustments to ensure constant product quality. 2 It is imperative that control systems are robust so that their utilization over a long period of time is fea- Correspondence to: Carl A. Anderson (Telephone: +412-396- 1102; Fax: +412-396-1608; E-mail: andersonca@duq.edu) Journal of Pharmaceutical Sciences © 2012 Wiley Periodicals, Inc. and the American Pharmacists Association sible. Efficient implementation of control systems re- quires well developed design of experiments (DOE), first-principle calculations, and rigorous statistical modeling. DOE is necessary to identify important fac- tors, calculate their effects on the response factors, and identify interactions between factors. 3–6 First- principle controls are desirable because they describe the major mechanisms by which the manufacturing process impacts product properties, reduce the dimen- sionality of subsequent DOE, account for external variability that cannot be eliminated, and potentially provide direct scale-up. Finally, statistical modeling is necessary to calculate the probability of success at all combinations within the measured knowledge space. This enables informed decision making within the boundaries of the knowledge space. The control system that was used in the current study has been described in detail previously, 7 and utilizes the environmental equivalency factor (EEF) as a first-principle variable. 8 This variable was shown to control the drying environment inside the fluid bed chamber despite substantial external environ- ment fluctuations by adjusting process parameters in JOURNAL OF PHARMACEUTICAL SCIENCES 1