Unit Dose Sampling: A Tale of two Thieves

Abstract

As a consequence of the United States vs. Barr Laboratories decision, pharmaceutical companies are now compelled to demonstrate the uniformity of unit dose samples of final powder blends. This investigation was initiated in response to a previously reported failed attempt to validate a process for lower strengths of a currently marketed product. This process has a long history of providing high-quality commercial tablets at the higher strength. The failure occurred because the unit dose samples of the final powder blend were indisputably subpotent. Interestingly, during the validation effort, 799 tablets of various strengths were assayed and exhibited outstanding content uniformity and potency. It was hypothesized that this failure was due to sampling bias which occurred when small (unit dose) samples were extracted with a thief from a static powder bed that was 7 orders of magnitude greater in size. The purpose of this two-part investigation was to test this hypothesis. In both of these studies, sieve analyses were conducted on samples collected from 100 kg of the final commercial blend. From these data the ratio of the coarse to fine fractions was calculated. This ratio is directly proportional to particle size and, for this product, related to the concentration of drug in the sample. Two different thieves (A and B) and a variety of sampling conditions were compared in Study 1 and the results suggest that the coarse-to-fine ratio: (a) decreases with sampling depth, (b) is generally larger for samples extracted with thief B than thief A, and (c) is larger for samples collected with a thief that is maintained in the vertical position than one held at an acute angle. The intent of Study 2, which focused exclusively on thief B, was to determine if the pronounced effect of sampling depth on the coarse-to-fine ratio was due to sampling bias or product segregation. The results of Study 2 demonstrated that coarser material was preferentially sampled from the top of the bed than from the bottom. These studies indicated that one thief used under different conditions and two thieves used under similar conditions can extract samples of different particle size from the same population. This implicates sampling bias and confirms that a thief is far from an ideal sampling device. We believe that the unit dose sampling requirement is more likely to impede the delivery of innovative pharmaceutical products to the marketplace than it is to enhance the quality of these products.