74. Are the data concerning analytical validation satisfactory? YES/NO If NO, recommended questions to the applicant appear in …………………………………………………………………… (eg page number below , or draft letter to the company on page …… ) Evaluation of analytical validation data The objective of the analytical procedure The analytical technique Item Data provided by applicant (very briefly) Acceptable or not? (add comments if necessary, & reasons if unacceptable) Is a chromatogram, spectrum or similar provided? Specificity Linearity Range Accuracy Precision: Repeatability Precision: Intermediate Precision: Reproducibility Detection limit Quantitation limit Robustness System suitability (if necessary) Data on the reference standard Other evaluator comments:
Notes de l'éditeur
NB From perspective of an evalr, not from perspective of an analyst experienced in HPLC.
Selectivity: Provide representative chromatograms with labelled peaks Suitable acceptance criteria might be: Capacity factor k’ > 1.0 Resolution >1.0 If ref imps not available, can prepare a suitable solution by forced degradation. If degradation >10%, may have to argue relevance. Diode array (suggested by ICH) may not be sensitive to low levels of imps, esp if chromophores similar When validating ID tests, remember to consider other actives on your site, as well as excipients
Linearity: Normally submit plot of detector response vs concentration, plus regression analysis Normally also determine accuracy and precision at extremes of range TGAL usually test across range of 50-150%, especially when test is to be used for CU and/or dissolution testing
Accuracy: When need to determine accuracy over a range of concentrations, normally 3 concentrations sufficient If mixture of excipients not available, may add known quantities of active and determine difference in results Care with reference substance. If a working standard (ie qualified wrt pharmacopoeial ref subst), bear in mind possibility that errors can accumulate. Care with hydration. Watch for time dependent variability, possibly associated with decomposition. If sample is filtered during preparation, specify type of filter and check for losses (soln of ref filtered vs not filtered)
‘ System’ repeatability: Aka repeatability between injections %CV can be of the order of 0.2%, particularly when automated FDA recommends min of 10 injections with %CV < 1.0% Many EP monographs prescribe %CV < 1% Acceptable %CV depends in part on the acceptance range for the assay in question Method repeatability: Aka repeatability of the complete method In practice, often estimated in conjunction with accuracy Means should be within + 2% of t/c Typically measured by: 6 replicates at 100% t/c OR 3 replicates of 3 concentrations %CV can be of the order of 0.4%, particularly when automated Acceptable %CV may be higher for: Impurities More complex matrices (such as creams) Microdose products
Intermediate precision: FDA uses the term ‘ruggedness’ with same meaning Although slide states means preferably within 2%, criteria are not laid down Acceptability is assessed on case by case basis
Reproducibility: Again: although slide states means preferably within 2%, criteria are not laid down Again: acceptability is assessed on case by case basis
LOD: LOD especially important for imps Submit typical chromatograms at LOD Std dev may be estimated from: Std dev of blank samples Std dev of intercept Residual std dev from regression analysis of calibration plot
LOQ: Again important for imps LOQ must be < acceptance criterion for the imp(s) !! Personally I would not want to rely on ‘visual’ evaluation for limit of quantitation. Submit typical chromatograms at LOQ As for LOD, std dev may be estimated from: Std dev of blank samples Std dev of intercept Residual std dev from regression analysis of calibration plot Confirm the LOQ by determining %CV at that value
Dot point 1: - If difference in assay >2%, describe attempts to improve. If little improvement gained, insert warnings into write-up of method. Dot point 2: - Duration and conditions of storage should represent most stressful situation likely to occur, eg during a 24hour auto-sampling run.