Reduced-Design Stability Studies
A new guidance covers bracketing and matrixing for the stability testing of drugs.
Erik Swain, Senior Editor
Pharmaceutical manufacturers interested in conducting "reduced design" stability studies now have a guidance showing them how to go about doing it. In these studies, not all package sizes, configurations, and fill volumes are evaluated at every single testing point in the study.
The document, "Q1D Bracketing and Matrixing for Stability Testing of New Drug Substances and Products," was written under the auspices of the International Conference on Harmonization and published in the United States
by FDA. It can be viewed at www. fda.gov/cder/guidance/4985fnl.PDF.
The guidance appears to encourage bracketing, which is testing only the extremes of certain design factors (such as container size) during the stability study, more than matrixing, which is testing different subsets of design factors at different time points.
The use of any reduced design should be justified, especially if the product has a complex delivery system with a large number of potential drug-device interactions. Such a design should be able to adequately predict the retest period or shelf life, and the manufacturer must consider the risk of having a shorter retest period or shelf life than what might have come from a full design.
Bracketing assumes that the stability of intermediate levels is represented by the stability of the extremes tested. The extremes would be tested at all time points, and the intermediates not at all. If a firm cannot demonstrate that the product strengths, container sizes, or fill volumes it wants to test are in fact the extremes, then bracketing should not be performed.
Bracketing can be used on studies of the same container closure system where either container size or fill varies while the other is constant. If both vary, it cannot be assumed that the largest and smallest containers represent the extremes of all packaging configurations. In that case, extremes should be selected by comparing other container characteristics such as wall thickness, closure geometry, and water vapor permeation rate. If justified, bracketing can be applied to studies for the same container with varying closures.
The shelf life for the intermediates should not exceed that for the least stable extreme.
Matrixing assumes that the stability of each subset of samples tested represents the stability of all samples at a given time point. Design factors that can be matrixed include strengths with identical or closely related formulations, batches made using the same process and equipment, and container sizes or fills in the same container closure system. Supporting data (such as similarities in moisture or light protection) are needed to matrix factors such as different closures.
In a matrixing design, each combination of factors should be tested to the same extent over the intended duration of the study and through the last time point before submission. However, the document states, "it may be difficult to achieve a complete balance in a design where time points are matrixed."
In general, a matrixing design is acceptable if the supporting data indicate predictable product stability. If the supporting data show moderate variability, a matrixing design should be statistically justified. If substantial variability is seen, matrixing should not be used.
A matrixing design on time points only is encouraged because it will "often have similar ability to that of a full design to detect differences in the rates of change among factors and to establish a reliable shelf life." But the document cautions that a matrixing design on factors other than time points "generally has less precision in shelf-life estimation and yields a shorter shelf life than the corresponding full design."
The document provides tables of sample matrixing designs: one for time points on a product with two strengths, one for time points on a product with three strengths, and one for time points, strengths, and container sizes for a product with three strengths and three container sizes.