Typical variability and evaluation of sources of variability in drug dissolution testing

To investigate variability in dissolution testing an international collaborative study was performed by 29 laboratories. Glibenclamide (glyburide) tablets were used in the investigation in which multipoint dissolution profiles were established using USP paddle apparatus. In contrast to a previous report, the variability of the glibenclamide dissolution data was significantly lower. Total variances (s(2)) were found to range from 18.34-44.18, Between Laboratory and Between Analyst variances (synthetic value) ranged from 12.9-38.7 and the Within Analyst variances ranged from 5.08-5.78. The dissolution profiles and corresponding variances obtained by laboratories with little or no experience in glibenclamide dissolution testing were similar to those obtained by more experienced laboratories, indicating that the test, especially when designed as multiple point dissolution testing, is sufficiently robust and capable of identifying differences in a manufacturing process or drug formulation. The smallest statistically detectable mean difference between two dissolution runs was calculated (95% CI) to be 7% for one analyst, or 5% if two analysts were to perform the dissolution tests.