Issue link: https://www.e-digitaleditions.com/i/1394390
www.biopharminternational.com July 2021 BioPharm International eBook 13 concentration is used in the robust- ness study. A formal risk assessment (ICH Q9) of all reagents, reagent lots, sample preparation, sample handling, method parameters, and sample hold times are evaluated for inclusion in the robustness study design. Software such as SAS or JMP are used to design the experiment with all of the main effects, t wo-factor interactions, and quadratic model terms included in the study design from the risk assessment. Three independent determinations are performed for each experimental run. Mean and SD of the n=3 determina- tions are calculated. The model is then fitted to the mean difference from the standard and SD results. The nominal best-set point or recipe for the method is defined, and clear operational lim- its are set from the robustness study. Limits are set where the method will be accurate (<10% of tolerance) and repeatable (<25% of tolerance) and documented in the method standard operating procedure. EVALUATING ACCEPTANCE CRITERIA When evaluating acceptability of an analytical method's error and variation, it should be viewed as a budget of error and how it inf luences out-of-specifica- tion (OOS) error rates. Method errors are random and normally distributed for the most part. Table VIII shows a typical recommended budget for the method and for the process. The variance of the errors is summative and can be viewed as a percentage of tolerance. Biopharmaceutical Analysis Method Validation Reportable result Method of analysis Acceptance criterion Difference between the positive control (standard) and the negative control and the 95% CI Means ANOVA t-test Must be statistically significant (p-value <0.05, two-sided test) Table IV. Study design 3: specificity identification 2x3. CI is confidence interval. ANOVA is analysis of variance. Reportable result Method of analysis Acceptance criterion Difference between the spiked and unspiked sample intercepts and 95% CI. Two-way ANOVA, concentration, spiked/unspiked and the concentration and spiked/unspiked interaction. ((Difference)/(USL-LSL))*100 Criterion is the same as accuracy, <10% of tolerance. Evaluate both the main effect of the spiked/ unspiked and the two-factor interaction if significant. Table V. Study design 4: specificity interference 3x3. CI is confidence interval. ANOVA is analysis of variance. USL is upper specification limit. LSL is lower specification limit. Reportable result Method of analysis Acceptance criterion Rate of degradation and the 95% CI from a linear or nonlinear fitted curve Linear or nonlinear fitted regression curve and all curve parameters and 95% CIs Must demonstrate a statistically significant (p-value <0.05, two- sided test) rate of degradation to be considered stability indicating Table VI. Study design 5: stability indicating 1x5. CI is confidence interval. Reportable result Method of analysis Acceptance criterion Robustness model equation for accuracy and repeatability. Set points for all method parameters evaluated. Operational limits for all method parameters evaluated. Multiple regression if all factors are continuous, or ANCOVA mixed model if both continuous and categorical factors are in the study design. Report only Table VII. Study design 6x: robustness. ANCOVA is analysis of covariance. Sources of error Recommended budget (% of tolerance) Method accuracy 10 Method repeatability 25 Method intermediate precision 30 Process mean delta from target 10 Process variation 25 Design margin 20% or more, depending on accuracy and the process mean delta from target Table VIII. Budget of analytical error. Figure 1. Assay evaluation calculator. FIGURE COURTESY OF THE AUTHOR.