Development and validation of an UPLC method for determination of content uniformity in low-dose solid drugs products using the design space approach

A simple and reproducible UPLC method was developed and validated for the quantitative analysis of finasteride in low-dose drug products. Method validation demonstrated the reliability and consistency of analytical results. Due to the regulatory requirements of pharmaceutical analysis in particular, evaluation of robustness is vital to predict how small variations in operating conditions affect the responses. Response surface methodology as an optimization technique was used to evaluate the robustness. For this, a central composite design was implemented around the nominal conditions. Statistical treatment of the responses (retention factor and drug concentrations expressed as percentage of label claim) showed that methanol content in mobile-phase and flow rate were the most influential factors. In the optimization process, the compromise decision support problem (cDSP) strategy was used. Construction of the robust domain from response-surfaces provided tolerance windows for the factors affecting the effectiveness of the method. The specified limits for the USP uniformity of dosage units assay (98.5–101.5%) and the purely experimental variations based on the repeatability test for center points (nominal conditions repetitions) were used as criteria to establish the tolerance windows, which allowed definition design space (DS) of analytical method. Thus, the acceptance criteria values (AV) proposed by the USP-uniformity of assay only depend on the sampling error. If the variation in the responses corresponded to approximately twice the repeatability standard deviation, individual values for parcentage label claim (%LC) response may lie outside the specified limits; this implies the data are not centered between the specified limits, and that this term plus the sampling error affects the AV value. To avoid this fact, the limits specified by the Uniformity of Dosage Form assay (i.e., 98.5–101.5%) must be taken into consideration to fix the tolerance windows for each factor. All these results were verified by the Monte Carlo simulation. In conclusion, the level of variability for different factors must be calculated for each case, and not arbitrary way, provided a variation is found higher than the repeatability for center points and secondly, the %LC response must lie inside the specified limits i.e., 98.5–101.5%. If not the UPLC method must be re-developed.