(T1130-10-58) Effective Product Life-Cycle Management Through Continued Process Verification (Stage 3) Using Statistical Software

Purpose: A case study is presented to demonstrate the implementation and benefits of a Stage 3 validation activity which includes collection, storage, trending, and statistical analysis of data related to an oral solid dosage (OSD) hard gelatin capsule product.  The outlined activities constitute a practical approach to a Stage 3 Continued Process Verification program. Good product stewardship and product life-cycle management require diligence in maintaining historical and real-time documentation of process performance.  These activities are part of a robust Stage 3 Continued Process Verification program for commercial products.  The Stage 3 validation phase was based upon a thorough risk assessment and control strategy defined in Stage 1 and qualified in Stage 2.   The Stage 3 program was defined and executed by monitoring critical data such as Critical Material Attributes (CMAs), Critical Process Parameters (CPPs) critical in-process tests/attributes, and Critical Quality Attributes (CQAs).

Methods: The CMA, CPPs, critical in-process tests/attributes, and CQAs for an oral solid dose product were defined. The manufacturing process consisted of a simple blend of drug substance and widely used excipients. The lubricated blend was encapsulated using dosing disk-type equipment. As a simple blend, the attributes of the raw materials such as particle size, density and moisture were selected to study. Assay and impurities of the drug substance directly affect those same qualities in the finished dosage form. Identified CMA data were collected and trended. The Critical Process Parameters were defined in Stage 1 and verified during Stage 2. These included the dispensed weights of raw materials, API pre-lubrication blend time, and the lubricant blend time. Critical in-process tests were defined as blend uniformity and filled capsule weights. Critical Quality Attributes were determined to be assay, water determination, related substances, Uniformity of Dosage (UofD), dissolution, and microbial evaluation. The defined critical data were stored using commercially available statistical software. The cause-and-effect relationships between CMAs, CPPs, Critical In-Process Attributes, and CQAs were studied to determine effectiveness of the materials, methods, and measurements control strategies. Corrective actions, or continuous improvements, were defined from the statistical analysis of these data.

Results: Basic statistical analysis of the critical data provided insight into stability and variability of the raw materials, in-process attributes, measurement systems, including sampling criteria, and product CQAs.  The continuous improvement benefits of the Stage 3 validation program were demonstrated for blend uniformity.  A process capability chart for one set of blend uniformity data indicated low capability (C_pk = 1.00) but also did not align with the finished product UofD data for the same batches (Figure 1). The finished product UofD data were used to determine if the blend uniformity data accurately reflected the distribution of drug substance throughout the batch. The variation contributed by the encapsulation process was also considered in this analysis. Theoretically, the variance of the BU data plus the variance contributed by the encapsulation process should be reflected in the finished product UofD data set. For the first three data sets, the variation of BU data added to the variance of the encapsulation process did not manifest in the variance of the UofD data set whereas the variance of BU data was greater than the variance of UofD data. The measurement process for BU, which includes sampling and analysis, was one potential source of the higher variance. A change was made to the sampling criteria which resulted in more capable process measurement and better alignment with the finished product Uniformity of Dosage data. As expected, the variance of the BU data was smaller than, or nearly equal to, that of the UofD data (Figure 2). The control of filled capsule weights was studied to determine the contribution to overall UofD variation. A X-Bar/R chart was prepared to determine the state of filled capsule weight control over time. This chart indicated the encapsulation process was stable with respect to filled capsule weight. A process capability analysis prepared with the same data showed adequate control of this critical in-process attribute (C_pk of 1.42).

Conclusion: The case study demonstrated the benefits of maintaining and analyzing historical and real-time process and product data. The data for BU testing indicated a process improvement was needed. A potential improvement was defined and implemented. BU data for subsequent batches demonstrated the effect of this change (Figure 3). Comparing the blend uniformity data to the finished product Uniformity of Dosage data provided assurance that the change implemented provided blend results more representative of the uniform distribution of drug substance throughout the finished product. This conclusion could be drawn in part due to the fact the variation contributed by the encapsulation process was stable and predictable. The ability to perform in-depth analysis of process data by using statistical software greatly enhanced the value of the critical data collected.