validation of analytical methods

Compendium of Analytical ProceduresCAP007/03Table of Contents1.2.3.4.4.14.24.34.44.54.5.14.5.24.5.34.64.74.7.14.7.24.7.34..4GLOSSARY 5System 5Assay (content or potency)...........................................................................................6Detection Limit (DL)..................................................................................................................6Quantitation Limit (QL).............................................................................................................7.....................................................................................................................................74.10 74.11 Range4.12 74.13 74.14 .74.15 74.16 Placebo (Blank Formulation).....................................................................................................74.17 Synthetic Formulation (Spiked Placebo)...................................................................................84.18 84.19 Coefficient of Variation (CV) or Relative Standard Deviation (RSD)......................................84.20 Peak Area Ratio (PAR)..............................................................................................................84.21 Working Concentration (Nominal Concentration).....................................................................TERS FOR VALIDATION ACCORDING TO 11Page 2 of 32

Compendium of Analytical ProceduresCAP007/..........................................................................................................................................12ACCURACY (RECOVERY).......................................................................................................21METHOD VALIDATION GUIDELINE SUMMARIES AND ACCEPTANCECRITERIA FOR HPLC 24EXAMPLE OF A GENERIC METHOD 32Page 3 of 32

Compendium of Analytical ProceduresCAP007/031. INTRODUCTIONThis document provides Company standards for validating analytical methods for drugsubstance and drug product. Its applicability and interpretation are defined in document will ensure that a consistent, acceptable approach is taken throughout theGroup companies in confirming that our analytical methods are suitable for their intendeduse. Concepts presented in this document may be generally applicable to other materialssuch as blends and raw materials and may be applied to other analytical techniques document will follow the guidelines given in ICH documents.2. SCOPEThis version of the method validation guideline has been written with specific references toHPLC assay, impurity and identity testing for drug substances and drug products. Theguidelines in this document may also be applied to other techniques. Additional validationguidelines are referenced in separate procedures covered in this document are guidelines and can be applied to drug substanceand drug product method validation exercises. In the early stages of development of a noveldrug or formulation, partial validation will usually be acceptable. These principles alsoapply to method re-validation, which may be required because of a change in formulation, amanufacturing process, in an analytical method, or when the regulatory standards acceptance criteria for the parameters described in sections 6 - 13 can be found insection 14.3. RESPONSIBILITYPerforming the appropriate level of validation for an analytical method is the directresponsibility of the validating scientist and their line management. However, a "one-team"approach to method development and validation is highly recommended. The "one-team"approach requires that the method development and validation group, communicate andinteract with the end-user group throughout the development, validation, reporting andtransfer of the 4 of 32

Compendium of Analytical ProceduresCAP007/034.

4.1GLOSSARY OF TERMSValidation TestsTests done to show that the method is scientifically sound and adequate for the intended use.4.2Analytical MethodThe analytical method refers to the way of performing the analysis. It should describe indetail the steps necessary to perform each analytical test. This may include but is not limitedto the sample, the reference standard and the reagents preparations, use of theequipment/apparatus, generation of the calibration curve, use of the formulae for thecalculation, etc.4.3Acceptance CriteriaCriteria which validation test results should meet for the method to be acceptable.4.4System Suitability TestsTests performed when a system is set up for a method to ensure that the system meets criteriarequired for reliable assay. System suitability tests are intended to be carried out each time asystem is set up for a particular analysis. They ensure conditions and hence results arereproducible both qualitatively and quantitatively.4.5SpecificityNote:This document refers to the term specificity, as this is now the preferred ICHterminology and should be used instead of the term icity is the ability to assess unequivocally the analyte in the presence of componentsthat may be expected to be present. Typically these components might include impurities,degradation products, excipients, matrix, etc. Lack of specificity of an individual analyticalprocedure may be compensated by other supporting analytical procedure(s). This definitionhas the following implications:4.5.1IdentificationTo ensure the identity of an analyte.4.5.2Impurity testsTo ensure that all the analytical procedures performed allow an accurate statement of thecontent of impurities of an analyte, i.e. related substances test, heavy metals, residualsolvents content, 5 of 32

Compendium of Analytical ProceduresCAP007/034.5.3Assay (content or potency)To provide an exact result which allows an accurate statement on the content or potency ofthe analyte in a sample.4.6AccuracyThe accuracy of an analytical procedure expresses the closeness of agreement between thevalue that is accepted either as a conventional true value or an accepted reference value andthe value found.4.7PrecisionThe precision of an analytical procedure expresses the closeness of agreement (degree ofscatter) between a series of measurements obtained from multiple sampling of the samehomogeneous sample under the prescribed conditions. However, if it is not possible toobtain a homogeneous sample it may be investigated using artificially prepared samples or asample precision of an analytical procedure is usually expressed as the coefficient of variation ofa series of ion may be considered at three levels: repeatability, intermediate precision andreproducibility.4.7.1RepeatabilityRepeatability expresses the precision of a homogeneous sample under the same operatingconditions over a short interval of time by one analyst using one instrument.4.7.2Intermediate precisionIntermediate precision expresses the precision of a homogeneous sample within a laboratoryon different days, different analysts and different equipment, possibly in the same laboratory.4.7.3ReproducibilityReproducibility expresses the precision between laboratories. Current ICH guidelinessuggest that reproducibility is usually determined by inter-laboratory collaborative study.(Reproducibility is not required for NDA and MAA filings and is only required forpharmacopoeial registration).4.8Detection Limit (DL)The Detection Limit of an individual analytical procedure is the lowest amount of analyte ina sample that can be reliably 6 of 32

Compendium of Analytical ProceduresCAP007/034.9Quantitation Limit (QL)The Quantitation Limit of an individual analytical procedure is the lowest amount of analytein a sample that can be quantitatively determined with suitable precision and accuracy. TheQuantitation Limit is a parameter of quantitative assays for low levels of compounds insample matrices, and is used particularly for the determination of impurities and/ordegradation products.4.10LinearityThe linearity of an analytical procedure is its ability (within a given range) to obtain testresults from an analytical response which are directly proportional to the concentration(amount) of analyte in the sample.4.11RangeThe range of an analytical procedure is the interval between the upper and lowerconcentration (amounts) of analyte in the sample (including these concentrations) for whichit has been demonstrated that the analytical procedure has a suitable level of precision,accuracy and linearity.4.12RobustnessThe robustness of an analytical procedure is a measure of its capacity to remain unaffectedby small, but deliberate variations in method parameters and provides an indication of itsreliability during normal usage. For a non-robust parameter, very tight limits possibly withno deviation allowed are required to control it.4.13ImpurityA compound not intended to be present. In a formulated product this may include residuesfrom the processing (e.g. solvents from film coating or granulation) as well as impuritiesfrom the drug substance.4.14Degradation ProductAn impurity resulting from degradation of the drug substance, which may include reactionproducts of the drug substance with excipients.4.15ExcipientMaterial other than the drug substance deliberately added to a formulated product.4.16Placebo (Blank Formulation)Contains all the components of the formulation except the analyte and related 7 of 32

Compendium of Analytical ProceduresCAP007/034.17Synthetic Formulation (Spiked Placebo)Contains all the components of the formulation including an accurately measured amount ofthe drug substance. In some cases it may be prepared by addition of a solution of the drugsubstance to an appropriate quantity of blank and synthetic formulations are prepared without being subjected to the processingused to make the formulated product.4.18Standard FormulationContains all the components of the formulation and is subjected to the processing used tomake the product, i.e. it is a sample of the formulated product. However, if used for theAccuracy test it should be specially selected or prepared to contain as precise an amount ofthe drug substance as possible.4.19Coefficient of Variation (CV) or Relative Standard Deviation (RSD)Is a measure of relative dispersion of data equal to the ratio of standard deviation (S.D.) tomean (M), usually expressed as a percentage:CV or RSD =

4.20SDx100MPeak Area Ratio (PAR)Is the area of the analyte peak divided by the sum of the areas of all the peaks of interest andexpressed as a percentage. The term is synonymous with normalised peak area (NPA)4.21Working Concentration (Nominal Concentration)The resulting sample assay concentration when the sample preparation procedure is followedas described in the relevant 8 of 32

Compendium of Analytical ProceduresCAP007/035. PARAMETERS FOR VALIDATION ACCORDING TO TEST METHODParameters requiring validation, according to development phase and test method type, aregiven in the table terSpecificityLinearityAccuracyRepeatabilityIntermediatePrecisionEquivalencyReproducibilityDetection LimitQuantitation LimitRobustnessStability InSolutionP1P3RTMCVABSSQN/AHPLCAssayP1P1P1P1P3TMCV/ABSCV/ABSRP1IdentityP1SQSQSQN/ATN/AN/AN/AN/AP1RelatedSubstancesP1P1P1P1P3TMP1P1RP1DissolutionP1P1P1P1P3TMN/AN/ARP1Non-ChromatographicP1P1P1P1P3TMCV/ABSCV/ABSRP1Required from Phase I onwards (proof of technology)Required from Phase III onwards (clinical studies)Required at time of Regulatory Submission (MAA/NDA)Required at time of method transferRequired for Monograph SubmissionRequired for Cleaning ValidationRequired for Absence DeterminationsRequired for Semi-Quantitative IdentificationNot ApplicablePage 9 of 32

Compendium of Analytical ProceduresCAP007/036. SPECIFICITYPrinciples - Specificity is usually accomplished by either demonstrating adequate separationof components, for example as in chromatography and/or by employing specific drug substance release assay methods may not need to be specific for drug relatedimpurities if separate methods are used to determine these impurities. Some drug productrelease assay methods may not need to be specific for impurities that are controlled in theinput drug substance if the levels of impurities do not increase during gh specificity against such impurities may not always be required, an appreciation ofthe degree of interference is required. Any achiral HPLC method is strictly “non-specific"for a chiral molecule but will still be considered “specific" if the enantiomer is determined bya second chiral s - The samples that are used to validate specificity should be as representative of“real" samples as possible and/or contain significant levels of potential interference. Forstability-indicating studies, samples should be stressed according to CAP032. Stressed oraged excipients are optimal for stability indicating drug product methods. If a stability-indicating method is being redeveloped and/or revalidated for a marketed product, expiredsamples and excipients may be used in place of or along with stressed s should be spiked with all available interferents at appropriate levels. Samples of thecrude drug substance or mother liquors may be of value when validating release assay andimpurity methodsProcedure - Samples of the analyte(s) and potential sources of interference may be injectedseparately and as mixtures onto the column at their “target” concentrations and run using thestandard method conditions in order to see where each elute. Samples that are most likely tohave interference present are run by the test method, and additional detection systems areemployed in order to detect any analyte peak non-homogeneity. It should be kept in mindthat impurities that have similar UV spectra to that of the analyte peak may not be detectedby this procedure. Also, even impurities having significantly different UV spectra will notbe detected if their retention times (chromatographic peak maxima) are the same as that ofthe main analyte the reasons stated above, a similar evaluation using another detection system such asMS, NMR, etc., is recommended in tandem with the chromatographic run. Alternatively,another chromatographic system with significantly different selectivity may be of value inorder to “challenge" the selectivity of the system under investigation. Another option is tocollect the analyte peak and evaluate the material collected spectroscopically or by anotherchromatographic system such as HPLC, GC, TLC, CE, SFC, etc., that has a large change inthe selectivity – Chromatograms and spectroscopic data of all the samples run should be used in theevaluation. Peak resolution should be calculated when 10 of 32

Compendium of Analytical ProceduresCAP007/037. LINEARITYPrinciples - The evaluation of linearity is most critical for major peaks since only very largedecreases in light reaching the UV detector (large absorbances resulting in very little lightreaching the detector) are likely to result in non-linear detector responses. For impurity limittest methods, a formal linearity study is not required. Impurity linearity experiments may nothave to be performed on every named impurity detectable by the method. This is anacceptable approach as long as those impurities evaluated are appropriate models (havesimilar integration properties) for those not evaluated, i.e., have similar sensitivity, peakshapes and are similarly resolved from other a calibration plot is not linear, every attempt should be made to ascertain the reason example, sample loading, insufficient light reaching the detector due to mobile phaseabsorbance, inappropriate detector or integrator time constant, data sampling rate or rangesettings, peak shape, peak resolution, chemical or physical interference etc., are parametersthat may be investigated in chromatographic analyses. If the active ingredient detectorresponse can not be adjusted so that it is linear up to at least 150% of the targetconcentration, then at least five points within the 50 to 120% range can be fact that the method was not linear up to the 150% level should be mentioned in theanalytical method and the addition of a linearity system suitability test should be would be extremely unusual for a theoretical non-linear detector response to analyteconcentration relationship to exist for absorbance values of up to at least 1.0, and sometimessignificantly higher, for HPLC assay methods of a simple pharmaceutical drug substance orproduct. The absorbance of the mobile phase at the analytical wavelength must beconsidered and can contribute to non-linearity especially at lower analyte absorbencies. Ifa non-linear method for the active ingredient must be used in order to achieve sufficientsensitivity for impurities, monitoring two wavelengths may offer a solution to the not, two separate methods should be considered for controlling the active ingredient andthe impurities. If all else fails, a suitable multi-level calibration and curve-fitting proceduremay be employed if this can be duplicated accurately by the end user s - The linearity of the detector response of the active ingredient is normallyevaluated by preparing a number of reasonably distributed standards ranging from 50 to150% of the target concentration. A minimum of five samples is typically used for thisevaluation. For impurities, the active ingredient (or active ingredient plus placebo for drugproducts) is typically spiked with each available impurity standard. Their concentration, inunits of percent of their upper specification, typically corresponds numerically to the activeingredient's percent target concentration, e.g., an amount of each impurity corresponding to120% of the upper specification limit would be spiked into the active ingredient at 120% ofits target concentration. If the lowest concentration of the impurities present in thesesamples is significantly greater than their reporting level, then further dilutions should beconsidered in order to bring the concentration of the impurities down to near the reportinglevel. Normally three solutions are prepared at each analyte concentration for samples inwhich the active ingredient is within the range of the method and two solutions are preparedfor concentrations outside of the method's 11 of 32

Compendium of Analytical ProceduresCAP007/03The sample preparations that are used for the linearity experiments (within the range of themethod) are normally the exact same solutions that are also used for the accuracy andrepeatability experiments (Sections 9 and 10). All analytes evaluated in the linearity studyare normally evaluated together in the same solution and therefore the same ure – The sample preparations described above are evaluated according to theconditions stated in the analytical method. Single injections of samples are normally usedwith modern HPLC injectors. With older equipment the samples may need to be injected atleast in – The detector responses for each analyte at each concentration is obtained. A linearregression analysis of the detector response verses amount of analyte applied to the columnor solution concentration is performed for each analyte. Individual data points or an averagepoint for each set of replicate injections of the same solution may be used for calculationsand plotting. The regression line should not be forced through the origin and a zero analyteconcentration point should not be part of the data equation of the best fit least squares line, coefficient of determination (R), slope of theline, Y intercept, and the residual sum of squares should be calculated and reported for eachplot. Other statistical treatments on the data may be applied as appropriate.8. RANGEPrinciples – The range indicates the concentrations of analytes over which data has beencollected which supports that the method will yield accurate results. The range normally justpertains to the major components such as the active ingredient or a preservative in a dosageform present at more than a trace gh accuracy and precision data may be collected at several levels for each availableimpurity, the “range" for each impurity is typically defined as the highest impurity levelutilised in the linearity study and that impurity's s, Procedure and Data – The range for the active ingredient in a drug substance isnormally demonstrated at 80, 100 and 120% of the target concentration. This same range orgreater is typical for drug products depending on the intended use of the method. The rangeshould normally include concentrations that are 10% lower than the lowest concentration and10% higher than the highest concentration expected to be encountered in the analysis. Thisoften corresponds to +/- 10% of the specification range for the active ingredient. Thus, awider concentration range would need to be demonstrated for a content uniformity or adissolution method as compared to the assay method. The validated range may be requiredto be extended to quite low concentrations for controlled release dissolution methods. Thesamples, procedures and data used for the linearity, accuracy and precision experiments aredescribed in their respective sections in this CAP. Normally the same samples and even thesame chromatographic run are used for all three sets of experiments for both the activeingredient and for impurities (see Section 14).Page 12 of 32

Compendium of Analytical ProceduresCAP007/039. ACCURACY (RECOVERY)Principles - Accuracy is a measure of getting the correct (or true) answer. In order toconsistently generate accurate results, an analytical method must be specific, linear,precise, sensitive, robust and possess good cy involves ensuring that (1) all of the analyte in the sample dissolves, (2) nothingwill prevent a representative portion of the analyte from reaching the detector, (3) nothing isextracted from the environment that will add to the analyte’s response, and (4) the detectorresponses (signal per unit amount placed on the column) of the analyte in the sample is thesame as that in the accuracy of the active ingredient from a drug substance is assumed to be 100.0%because the physical and chemical environment of the analyte in the sample is essentially thesame as that of the standard used in the analytical method. This of course assumes that thestandard and sample preparation procedure (including solvents, etc.) are the same. Anysmall differences between the sample and standard due to the level of impurities present ineach can be assumed to be insignificant. A safety factor should always be built into thesample preparation procedure (e.g., time or conditions required to dissolve the sample) dueto any differences in the physical states such as; particle size, degree of solvation, crystallineform, etc. Thus, because the accuracy of the active ingredient in a drug substance methodcan be inferred, a method is considered accurate (in its true full meaning) once specificity,linearity, precision and robustness have been for a drug substance method, the accuracy for the active ingredient in a dosage formassay method can not be inferred once the specificity, linearity, precision and robustnesshave been demonstrated. This is because the chemical and physical environment of theanalyte in the sample and/or the sample solution may be significantly different from that ofthe standard. Therefore the “accuracy" of the analyte from the dosage form and from thestandard may be significantly ties may or may not “behave" the same way when in the presence of the other samplecomponents, and therefore accuracy should be evaluated. The accuracy of impurities insamples can be determined in drug substance or drug product samples by spiking thesesamples with known levels of impurities and quantitating their increase against impurityreference standards containing no other sample components. The accuracy of impurities indrug substance samples is typically very close to 100%. Any accuracy of less than 100% forimpurities in drug substances are typically more associated with interactions with glassware,tubing, surface materials, etc., than the chemical/physical interference’s also affectingimpurities in drug accuracy of each impurity may not be needed if the impurity level is calculated on anarea percent basis, versus the active ingredient, and the impurity specification wasestablished on that basis using the same analytical system. In this case, the absolute accuracyof the impurity would not have to be known. The accuracy is assumed to be consistent fromsample to sample and over time for a given analytical system. This consistent accuracy willallow the impurity to be controlled in a manner consistent with the intent of the 13 of 32

Compendium of Analytical ProceduresCAP007/03An accuracy factor plus a relative response (detector) factor that is specific to each impurity-active ingredient pair and to the method conditions (analytical system) can be used in orderto obtain a true weight addition to accuracy experiments, consideration may be given to comparison of the resultsfrom the method undergoing validation with those obtained from an alternate validated testprocedure. It is unlikely however that another appropriate validated quantitative method isavailable for the analytes required to be controlled in that specific sample matrix. This isespecially true for unmarketed drug substances and sActive Ingredient in a Drug Product - Classical accuracy samples are prepared by spiking aprecisely known quantity of the active ingredient into a placebo typically over the range of 80to 120% of the target concentration. It may be appropriate to alter this range depending uponthe specific purpose of the method, e.g., content uniformity (typically 70 - 130%) ordissolution (typically 50-150%). The placebo should also contain the available knownpotential impurities near their upper specification limits. Often all excipients are mixedtogether in the proper proportion by the analyst if a placebo sample is not available. Whenpossible each component should be quantitatively added together in a way to simulate thephysical state of “real” samples. For example, if possible, solid components should be addedtogether for solid dosage forms instead of adding solutions of each component together. It isoften impractical to accurately add impurity standards in the solid state because of the smallamounts required. Normally solutions of impurity standards in volatile solvents are used forpreparation of accuracy samples. Three levels of the active ingredient are prepared at aboutthe 80, 100 and 120% of the target level. Three preparations are made at each concentrationlevel. Standard solutions of the active ingredient should be prepared as described in theanalytical ed aged and fresh drug product samples can also be used in order to assess theaccuracy of the method by using exhaustive extraction procedures as described in the next(Procedure) reason why this may add value is that some manufacturing processes of some dosageforms (e.g., tablet compression) can cause physical conditions that affect analyte extractioncompared to when the components are simply added together. The challenge to thisapproach is that it is difficult to know the exact analyte content of any batch of formulatedproduct. Therefore, it may be hard to determine if the similarity in the assay values for theexhaustively extracted samples and the samples extracted according to the method is due toboth methods fully extracting the analyte or if both have the same degree of “non-accuracy".Determining the differences in accuracy between the standard and exhaustive extractionprocedures will be hindered by any differences in the analyte content between the two sets ofsamples used in the experiment. This can be somewhat overcome by using samples that arehighly uniform (refer to batch content uniformity data) and evaluating a larger number ofsamples per each procedure. A similar approach is to perform an analysis on the samplesolution after normal sample preparation, and then perform more vigorous techniques on thatsame solution in order to try to extract more 14 of 32

Compendium of Analytical ProceduresCAP007/03This approach takes care of the sample non-uniformity issue but any changes to thesample volume must be carefully considered in the assay ties in a Drug Product – The available impurities spiked into the placebo describedabove should be done so quantitatively. These impurity standards that were spiked into theplacebo near their upper specification levels, for the accuracy study for the active ingredient,form the basis of the impurity accuracy study. Impurity standard solutions near their upperspecification levels that do not contain the active ingredient or product excipients arenormally used as the reference Ingredient in a Drug Substance – As previously stated, the accuracy of the activeingredient in a drug substance sample is assumed to be 100.0% if the sample and standardare treated in the same manner. Therefore accuracy only has to be demonstrated forimpurities in drug substances. This may include residual solvents, which are normallyassayed by GC ties in a drug substance - Normally drug substance samples are quantitatively spikedwith available specified impurities at their upper specification level. Impurity standardsolutions at levels near their upper specification levels that do not contain the activeingredient are normally used as the reference ureDrug Product - The placebo (containing the available controlled impurities) and that placebomix spiked with active ingredient samples are evaluated versus an active ingredient and animpurity reference standard solution by application of the analytical method underconsideration. If drug product samples are spiked instead of placebos, then these spiked andunspiked samples are evaluated by application of the normal analytical method conditions. Itis important to determine the impurity levels in the unspiked drug product or drug substancesample as accurately as possible so that the accuracy of the impurities can be experiments can be supplemented by comparing results of the amount of activeingredient or impurity extracted from typical fresh and aged drug product samples, such as atablet or capsule, by the conditions specified in the test method verses results obtained byextraction using exhaustive example, increasing extraction time, increasing solvent strength, adding an all aqueouspre-disintegration step, adding a sonication step, adding a soaking step and then additionalshaking, using longer and more vigorous homogeniser settings, etc. These exhaustiveconditions may not be practical to use on a routine basis, but for these experiments, oftenprovide a good reference for complete extraction of analytes from real Substance - Spiked and unspiked drug substance samples and the impurity standardsolutions (with no active ingredient present) should be run according to the method 15 of 32

Compendium of Analytical ProceduresCAP007/03Data - Accuracy is usually expressed in terms of the increase or decrease in the analytecontent measured by the method relative to the amount of analyte that was actually spikedinto the placebo, drug substance or drug product sample.æRs-RuPercent accuracy (recovery)

=ççRaèWhereö÷÷×100øRs = assay result of the spiked sample/placebo vs. the standard solutionRu = assay result of the unspiked sample/placebo vs. the standard solutionRa = actual total amount of analyte in the spiked sample/placeboIn cases where exhaustive extraction is employed, the assay value determined using thenormal method conditions is compared to the assay value determined using the exhaustiveæRnPercent accuracy (recovery)

=

ççRèeextraction :ö÷÷×100øRn is the assay result generated under normal conditionsRe is the assay result generated under exhaustive conditions10. PRECISIONPrinciples - The precision of methods can be evaluated for three circumstances. Therepeatability, intermediate precision and reproducibility If the reproducibility of a method isdetermined, it is usually not necessary to determine the intermediate precision sincereproducibility is normally a greater challenge, and incorporates or satisfies all of therequirements of intermediate abilitySamples - It is important to use homogeneous samples in all precision studies so that theresults are not significantly influenced by the sample, rather than the precision limitations ofthe ingredient – Three preparations are normally prepared at each the upper, middle andlower range concentrations, e.g., 80, 100 and 120% of the target concentration for both drugsubstances and 16 of 32

Compendium of Analytical ProceduresCAP007/03The samples and data used for the repeatability study are normally the same used for theaccuracy and linearity experiment. Alternatively six preparations at 100% of targetconcentration may be ties - Three preparations at each the upper, middle and lower range , 80, 100 and 120% of the upper specification levels are normally prepared, for impuritiesin both drug substances and products. Alternatively six preparations at 100% of targetconcentration may be used. It may be difficult to obtain a standard for some impurities or tofind homogeneous samples that contain all of the “representative" impurities at levels nearthe upper specification levels. All of the “representative" impurities means that there are noother impurities to be controlled by that method having significantly different peakresolution, sensitivity, shape, etc. Samples may be accurately spiked in order to achieve theappropriate impurity levels when an impurity standard is available. The samples and dataused for the repeatability study are normally the same used for the accuracy and ure – All solutions should be evaluated using the conditions stated in the methodbeing evaluated vs. the same reference standard(s). Each solution is normally injected intriplicate. The active ingredient and the impurities should be evaluated in the samechromatographic run for samples, i.e., should be present in the same sample solution, asappropriate. Standard solutions should only contain either the active ingredient or – Typically nine results are considered for each analyte. These are normally the sameresults that were generated for the accuracy and linearity experiments. Injections are madeon three preparations each at the 80, 100 and 120% of target levels for actives and 80, 100and 120% of the upper specification limits for impurities. The assay values and the detectorresponse per unit analyte concentration or per amount of sample placed on the column arecalculated for each analyte. The %RSD of these nine values (one value for each samplepreparation) for each analyte is also ediate Precision and ReproducibilitySamples - It is important to use homogeneous samples in all precision studies so that theresults are not significantly influenced by the sample, rather than the precision limitations ofthe method. A minimum of six sample preparations are normally prepared and analysed byeach analyst participating in the study. For impurity methods, it may be difficult to findhomogeneous samples that contain all of the representative impurities at levels near theirupper specification limits. All of the representative impurities means that there are no otherimpurities to be controlled by that method having significantly different peak resolution,sensitivity, shape, etc. Sample solutions may be accurately spiked in order to achieve theappropriate impurity levels. When different sets of data need to be obtained for a spikedsample, i.e., the same spiked sample is to be tested in multiple laboratories, portions of thesame impurity mixture solution should be spiked into the same sample batch. Allparticipating laboratories throughout the study should also use the same reference 17 of 32

Compendium of Analytical ProceduresCAP007/03Procedure - Each analyst should perform their analysis as independently as possible from theothers, using their own reagents, standard solutions, equipment, column, and instrumentwhen possible. Normally a minimum of six replicate assay evaluations are made on eachsample (or spiked sample) according to the conditions in the analytical – The assay value for each sample preparation, the overall average assay value for eachbatch of sample (or spiked sample) and the %RSD of the assay values based on eachreplicate sample preparation is calculated for each analyte.11. DETECTION LIMITPrinciples - The detection limit is the lowest concentration or amount of analyte in a samplethat can be detected, but not necessarily accurately or precisely quantified, under the statedexperimental conditions. This information is only relevant for impurity or trace level assays;therefore, a detection limit for the active ingredient in a drug substance or drug productmethod is normally not required. Most commonly, the detection limit is defined as theamount or concentration of analyte that gives a signal-to-noise ratio of 3:1. This only appliesto methods that exhibit baseline noise such as most chromatographic are also statistical approaches that can be used to estimate detection detection limit of each impurity will need to be estimated since the sensitivity of eachcompound may vary from one to another based on; their specific spectral characteristics,chemical state (e.g., ionised or complexed or not), peak shape, the baseline noise at theelution time of each impurity, etc. The equipment model and condition, reagent quality, also affect the detection limit over detection limit can be expressed as an area percent. In this case, the relative responsefactor, and even the structure, does not have to be known. The area of an impurity thatproduces a signal intensity three times that of the baseline noise level is compared to theactive ingredient peak area when at its nominal concentration. The equation is Ai/Au X 100where Ai is the impurity peak area when its peak intensity is three times that of the baselinenoise level and Au is the active ingredient peak area when at its nominal detection limit of impurities in a drug substance or product can also be expressed as theconcentration of the impurity in a sample in terms of a weight percent relative to the activeingredient. For example if 50 µg of that active ingredient is placed onto the column and theimpurity can be detected down to the 0.005 µg level, then the detection limit for thatimpurity for that method is:0.005 µg×100=0.01%ww50 µgA relative response factor (RRF) can be used in order to obtain a weight percent detectionlimit from an area percent. This factor is specific to each impurity and active ingredient pair,and the instrument and method parameters. The robustness of the relative response factor isdependent on the UV characteristics of the compounds involved, and the selected instrumentand method 18 of 32

Compendium of Analytical ProceduresCAP007/03If the RRF is defined as the chromatographic peak response of the active ingredient per unitconcentration divided by the chromatographic peak response of the impurity per unitconcentration, then the area percent is multiplied by the response factor in order to obtain theweight percent Relative Response Factor (RRF) should be used if a more accurate estimatethan the area percent figure for a specific impurity is needed. It is recommended that thisapproach is taken only if the RRF of the impurity is significantly different from unity ( outside the range of 0.8 to 1.2). Response factors of less than 0.2 or more than 5 shouldnot be used. If the difference between the response of an impurity and that of the substancebeing examined is outside these limits, a different method of determination, such as adifferent detection wavelength, or a different detection technique (e.g., fluorescence), shouldbe considered. This will normally apply to UV responses, and is applicable only at aspecified wavelength. When determining RRF’s, the actual purity of the main componentand the impurity in question must also be considered. The method to be used forcalculating and using a RRF is:RRF=Specific Response for Main ComponentSpecific Response for ImpurityThis RRF factor should always be multiplied by the area percent figure of the impurity inquestion to give the equivalent %w/w s – A sample should be prepared that either contains no, or known very low levels ofthe impurities under investigation. A portion of this sample should be spiked with knownlevels of each controlled impurity so that the final impurity level is ideally in-between thedetection and quantitation limits. If any impurity is present at a level significantly outsidethat range, then it should be re-injected at a more appropriate level. Samples should not beevaluated having the impurity levels much higher than the quantitation limit and the resultsextrapolated to significantly lower detection or quantitation limit levels. If an impuritystandard is not available, then another representative impurity or a diluted sample of theactive ingredient may be used as a model for that ure –Injections of unspiked and spiked sample and any impurity standard solutionsshould be evaluated under the conditions stated in the method. The attenuation of the systemshould be adjusted so that the chromatograms contain a measurable amount of tration adjustments should be made for impurities present at levels that aresignificantly different from the reporting level or quantitation – For weight percent detection limits, the levels of each impurity in the spiked andunspiked samples need to be determined. The average or median baseline noise level at theretention time of each impurity needs to be obtained. Impurity standards or relative responsefactors are required to know the weight percent impurity levels in samples. The peak areaper unit amount or concentration for each impurity standard is calculated. The amount orconcentration of the impurity that would give a detector signal three times the noise level isthen calculated. That amount/concentration of impurity is then compared to the nominalamount/concentration of the active ingredient in the drug substance or drug product 19 of 32

Compendium of Analytical ProceduresCAP007/03For area percent detection limits, the peak area corresponding to a peak that has a heightthree times that of the baseline noise needs to be calculated. That impurity peak area is thenreported as a percent of the nominal peak area of the active ingredient in the drug substanceor drug tical Method of DeterminationThe Limit of Detection can also be obtained statistically from the linearity validation dataand using the following equation;DL =

3 Sy/xbWhere b is the slope of the Linearity correlation /x=å(yi−y)2n−2n is the number of data

is the observed peak area/height.y is the area calculated from Linearity correlation detection limit of an impurity should be reported in the same units as that given in thedrug substance or drug product specification. That will normally either be a weight or areapercent versus the active ingredient. If a reference standard of an impurity is not available,then the detection limit of another “representative” impurity or of the active ingredient maybe used as a model for that impurity.A pragmatic or visual approach to estimate the detection limit of an impurity would be tosimply establish, by running very low amounts of the analyte, the minimum amount that canbe reliably detected. The visual approach is normally used for non-instrumental test methodssuch as TLC employing visual either the detection limit or the quantitation limit (see Section 12) is estimated then theother limit may be calculable from the same data.12. QUANTITATION LIMITPrinciples - The quantitation limit is the lowest concentration or amount of analyte in asample that can be determined with acceptable precision and accuracy under the statedexperimental conditions. The quantitation limit is usually defined as the amount orconcentration of an analyte giving a signal-to-noise ratio of 10: 20 of 32

Compendium of Analytical ProceduresCAP007/03The quantitation limit is estimated by considerations that are directly analogous to thosestated in Section 11 for the detection limit, so these will not be repeated in this either the detection limit or the quantitation limit is estimated then the other limit maybe calculable from the same s – The same samples that are used to estimate the detection limit can also be used toestimate the quantitation ure – The same chromatographic runs used to estimate the detection limit can also beused to estimate the quantitation - The same calculations and data that are used to estimate the detection limit can also bealso used to estimate the quantitation limit. The only difference involves the quantitationlimit being defined as the amount or concentration of analyte that produces a 10:1 signal tonoise ratio instead of a 3:1 tical Method of DeterminationThe Limit of Quantitation can also be obtained statistically from the linearity validation dataand using the following equation;10 Sy/xQL =

bNote: For key, see section 1113. ROBUSTNESSRobustness should be designed and built into each test method when it is developed. Therobustness of a method can be investigated by making deliberate small changes to relevantparameters in the testing procedure and then determining the effect on the resolution of peakpairs and on the quantitative results. Changes in the quantitative results can be caused by achange in how the analyte peaks are being cy and/or detector sensitivity differences between the analyte present in the standardand the sample preparations can also affect assay results. The acceptance criteria should beset so that after any reasonable change in any of the method parameters is made, no adverseeffect on the ability to accurately integrate peaks or on the quantitative results should bedetected.A lot of work should go into designing and building robustness into the method during themethod development phase. In the validation phase, the robustness is being confirmed afterall of the test method parameters have been optimised and selected. It should not benecessary to repeat experiments on every one of the method parameters. Scientificjudgement can be used to eliminate non-value repetition if the robustness of the final methodparameter can be properly documented by work that was performed in the methoddevelopment 21 of 32

Compendium of Analytical ProceduresCAP007/03Typically, only the most sensitive parameters for a particular method will be identified by thedevelopment laboratory to be included in the post-development validation study. The use ofexperimental design, e.g. involving fractional factorial techniques, is recommended. If it isnot possible to build robustness into a given parameter, then the non-robustness of thatparameter should be documented in the analytical method and acceptable operating limitsprovided for that non-robust s – Standards and samples should be prepared as stated in the analytical methodunless otherwise directed. For a given product formulation, only one dosage strength needsto be evaluated if the final analyte solution concentration is similar, e.g., ± 10%. Forimpurity methods, samples should be spiked if necessary in order to bring each impuritylevel close to its upper specification limit. If impurity standards are not available, samplescan be stressed to increase the level of degradation products. Crude drug substance ormother liquor samples can be used or spiked into “normal" samples in order to achievehigher levels of process s of the highest and lowest strength of each formulation are usually evaluated for thereproducibility portion of the precision study (Section 10) which can also be used to evaluatethe overall robustness of the ure - Some parameters that should be considered when evaluating the robustness of anHPLC assay method and typical amounts that these parameters may be varied are listedbelow. Since all of the appropriate parameters below will have been evaluated andoptimised during the method development stage, only a few of the most critical parametersfor the particular method under study need to be included in the formal method validationexercise. The quantitative range listed in the table for varying any specific parameter isprovided as a generic guideline actual amount that each parameter is varied from that specified in the method will beselected based on the intended use and requirements of that particular method. The typicalvariation percentages in the table regarding mobile phase, gradient and solvent changes are interms of absolute percentages. These absolute variances will probably be less and lessappropriate when dealing with methods that contain any of the components approaching thezero and/or one hundred percent levels. Thus, smaller absolute percentage changes mayneed to be selected as appropriate for some parameters listed in the table terInjection volumeWavelengthSample extraction timeStationary phaseTypical Variation± 25%± 4 nm± 50%3 different lots and/or “equivalent”manufacturersPage 22 of 32

Compendium of Analytical ProceduresCAP007/03ParameterMobile phaseorganicinorganicmodifierpHSample solventColumn temperatureFilter studySolution stabilityTypical Variation± 5%± 10%± 10%± 0.2 units± 5% organic and inorganic± 5 degrees C3 lots and/or “equivalent" manufacturersstore at room temperature and at refrigeratedconditions for a minimum of 4 days forsample and 2 weeks for standard solutionsrepresentative models in laboratoryEquipment/instrumentData – When evaluating most parameters, quantitative sample results are obtained underboth the modified and the standard conditions. These assay values must be obtained fromsample and reference standard data that were obtained from chromatographic runs usingidentical conditions (ideally from the same chromatographic run). The robustness of thetest method is usually evaluated by comparing the assay value obtained under the modifiedconditions to those obtained under standard conditions, i.e., those stated in the method. Thepercent difference in results obtained under modified conditions versus standard conditionscan be calculated by:Rmx100RsWhere: Rm is the assay result obtained using the modified conditionRs is the assay result obtained under the standard resolution between peak pairs obtained under both standard and modified conditionsshould also be calculated when quantitative information might offer value over thequalitative information gained by visually examining the acceptance criteria may not always be appropriate for all method parameters optimal robustness is not obtainable for some method parameter, then good scientificjudgement must be used in order to predict if that parameter can routinely be controlled tothe degree that is required to consistently produce accurate results. If a given parameter isnot robust enough to consistently generate accurate results, then the method is probably notsuitable for its intended use. This method would then need re-development work with focuson the non-robust parameter and the other parameters that may have an affect on the non-robust one. Once redeveloped, the method must then be re-assessed. Information must beprovided in the method validation report and in the analytical method describing thedegree of control that is required for any non-robust 23 of 32

Compendium of Analytical ProceduresCAP007/0314. METHOD VALIDATION GUIDELINE SUMMARIES AND ACCEPTANCECRITERIA FOR HPLC ASSAY METHODSThe guidelines contained in this document for HPLC assay methods are summarised in thetables terReleaseSpecificity - UV diodearray and an , MSwhich is normallyperformed during themethod developmentphaseResolution of processimpurity peaks fromactive tration ofpeak SubstanceStabilityImpuritiesLinearity – threepreparations at each ofthe 80, 100 and 120%levels, (twopreparations outside the80 – 120% range istypical)Range – 80 – 120%Accuracy (Recovery)– three preparations ateach levelResolution from the active andResolution ofdegradation productother impurity from tration ofpeak trated resolution of compounds from each other where appropriate bychromatographic separation (preferably R

≥ 1.5), or by the detection . 1.0 % interference for the active and max. 20% interference forimpurities. Peak purity value

≥999 (or the equivalent for systems other thanAgilent)50, 80 100, 120, 150% of targetMinimum five solutions reporting level up to 150%of the upper impurityspecification ination of Correlation coefficientDetermination of Correlation≥0.999(R), and the intercept is within +/-≥0.990and the intercept is +/-2.0% of the active response at its target20% of the response of thatconcentration, plot of residuals indicatesimpurity present at its ication limit, residualsindicates on acceptable linearity, accuracy andBased on acceptable linearity,repeatability resultsaccuracy and repeatabilityresultsImplied after specificity, linearity andAvailable process impuritiesand degradation productsrepeatability have been successfullyspiked into trated.(80%, 100% & 120% of theimpurity upper specificationlimit)N/AThe average of each of thethree levels is within +/- 20% ofthe actual level or see Table 1Page 24 of 32

Compendium of Analytical ProceduresCAP007/03ParameterReleasePrecision -Repeatability –threepreparations at eachlevel (80, 100, 120%)ofthe target concentrationor six preparations at100% Intermediate Precision& Reproducibility - (Reproducibility maybe performed in lieu ofIntermediate precision)Drug SubstanceStabilityImpuritiesReport individual impurityrepeatability based on the%RSD of detector responsesper unit concentration for the 9sample

≤20%Same runs used for the activeassay. Samples must containsignificant levels ofdegradation products andprocess RSD of the 6 assay valuesobtained on each impurity≤20% for each condition/lab,individual lab mean differsfrom the referencecondition/lab mean by

≤20% orwithin the ranges in Table ICHTarget 0.02%/0.05%.Report repeatability based on the overall%RSD of detector responses per unitconcentration for the 9 sample 6 preparation of 100% reportrepeatability based on the %RSD of these

≤1.0%.Minimum 6 replicate assays (samplepreparations) of the active for each %RSD of the 6 average assay valuesobtained for each sample preparation iscalculatedThe RSD for each batch

≤1.0% of the 6 assayvalues obtained on each condition/lab,individual lab mean differs from referenceconditions/lab by

≤1.0%.(Alternative: Statistically valid significancetesting )N/AN/ADL/QLTable 1Observed Impurity Level≤ 0.2%0.2% to 1.0%> 1.0TotalAgreement Acceptance Criteria± 0.05% absolute± 20% relative± 10% relative± 20% relativePage 25 of 32

Compendium of Analytical ProceduresCAP007/03ParameterReleaseRobustness – standardsolution stability -(Normally performed inmethod developmentstage)Sample solutionstability -(Normally performed inmethod developmentstage)Chromatographyparameters -(Most criticalparameters re-examinedin validation stage)Drug SubstanceStabilityImpuritiesThe solution stability experiments should be performed on solutions at periodicintervals up to at least two weeks for standard solutions or until the solutions falloutside of the acceptance criteria. Max. 0.5% loss calculated from % increase in+/- 10% of originalimpurities after defined time. Confirm by e vs. fresh appropriate, periodic intervals, confirmation vs. fresh standard+/- 10% of originalOnly required if the sample solvent is different standard solvent. Max. 0.5% loss calculatedfrom % increase in impurities after defined time(target min. 4 days).∆ Column, temperature, gradient, organic, pH, ionic strength, λ, etc., asappropriate+/- 20% of reference+/- 1.5% of reference condition, any non-robustness must be either corrected or ion or see Table,any non-robustnesseithercorrected/ CAP015 - System Suitability RequirementsSystem suitability -Page 26 of 32

Compendium of Analytical ProceduresCAP007/03ParameterReleaseSpecificity - UV diodearray and an , MS whichis normally performedduring the methoddevelopment tion of placebo andmanufacturing impuritypeaks from active tration of ProductStabilityImpuritiesLinearity – threepreparations at the 80,100 and 120% levels,(two preparations at eachlevel outside the 80 –120% range is typical)Range – 80 –120%Accuracy (Recovery) –three preparations at eachlevel (80, 100, 120%)Resolution from the active,Resolution of agedplacebo and other impurityplacebo ation productpeaks from activepeak. Demonstrationof peak trated resolution of compounds from each other where appropriate bychromatographic separation (R≥1.5), or by the detection method, erence of 1.5% for the active and max. 20% for purity value

≥999 (or the equivalent for systems other than Agilent).50, 80, 100, 120, 150% of target m five solutionsfrom the reporting level(generally 0.05%) up to150% of the impuritiesupper spec ination of Correlation coefficient

≥0.999(R),Determination ofand the intercept is within +/- 2.0% of the activeCorrelation

≥0.990, andresponse at it target concentration, plot of residualsintercept +/-20% of theindicates se of that impuritypresent at its upperspecification limit, on acceptable linearity, accuracy andBased on acceptablerepeatability results. (linearity and QL )linearity, accuracy andrepeatability results.(linearity and QL forimpurities)Active recovered fromActive recovered fromProcess and degradationimpurities spiked intoplacebo and availableplacebo andactive / excipient mixprocess impurities. available degradation(80%, 100% &120% ofproductsupper impurity limit).98.0% to 102.0% at each98.0% to 102.0% at The average result +/- 20% of theactual level or see Table repeatability based on mean %RSD of 3 levels (9 samples) at the 80%,100% &120% target concentration for the active and at 80, 100 and 120% of theupper specification level for

≤2.0%RSD

≤20%Precision -Repeatability - threepreparations at each levelPage 27 of 32

Compendium of Analytical ProceduresCAP007/03ParameterReleaseIntermediate Precision& Reproducibility - (Reproducibility maybe performed in lieu ofIntermediate precision.)Drug ProductStabilityImpuritiesSame runs used for theactive assay. Samplesmust containsignificant levels ofdegradation productsand process RSD of the 6 assayvalues obtained oneach impurity

≤2.0% for each lab, individual lab meandiffers from reference condition/lab by

≤2.0%.(Alternative: Statistically valid significance testing)( 1.0TotalAgreement Acceptance Criteria± 0.05% absolute± 20% relative± 10% relative± 20% relativePage 28 of 32

Compendium of Analytical ProceduresCAP007/03ParameterReleaseRobustness – standardsolution stability -(normally performed inmethod developmentstage)Drug ProductStabilityImpuritiesSample solutionstability - (normallyperformed in methoddevelopment stage)Analyte extraction -(performed indevelopment stage) SeeCAP053 for automatedpreparationsfilters - (performed indevelopment stage)Chromatographyparameters -(most criticalparameters re-examinedin validation stage)System suitabilityThe solution stability experiments should be performed on solutions at periodicintervals up to at least two weeks for standard solutions or until the solutions falloutside of the acceptance criteria.0.05 absolute or +/-Max. 1.0% loss calculated from % increase in10% of original valueimpurities after defined time. Confirm by assayingwhichever is the higheranalyte vs. fresh t appropriate periodic intervals, confirm assay vs. fresh standard.≤1.0% loss calculated from % increase in impurities0.05 absolute or +/-10% of original valueafter defined time (target min. 4 days).whichever is the higherExtraction profile (detector response vs. time), homogeniser times, speeds, etc.+/- 2.0% of referenceActive and impurity recoveries, extractives, flush volumes+/- 1.0% for assay methods, +/- 2.0% for dissolution and CU methods∆ Columns, temperature, gradient, organic, pH, ionic strength, λ, etc., asappropriate+/- 2.0%, any non-robustness either corrected or+/- 20% or see Table, non-robustnesseither corrected CAP015 – System Suitability RequirementsPage 29 of 32

Compendium of Analytical ProceduresCAP007/0315. EXAMPLE OF A GENERIC METHOD VALIDATION PROTOCOLAn example of a typical method validation consisting of a drug substance and two dosageforms will be described below. Dosage form A has three strengths (25, 50 and 100 mg) anddosage form B has two strengths (50 and 100 mg). The formulation for product A is thesame and only the tablet weight is changed in order to achieve the different productstrengths. For product B, the formulations of the two strengths are different. The activeingredient concentration of the sample solutions is the same for all dosage strengths. Thegeneric target reporting level of 0.05% for impurities will be used in this SubstanceSpecificity – The active peak of an aged sample is evaluated by UV diode arrayspectroscopy and an orthogonal technique for peak purity. An impurity mix comprised of allavailable impurities present at their upper specification level and a sample spiked with theimpurity mix are also run to demonstrate cy – A mixture of the active ingredient and impurities “both" present at 80, 100 and120% of their target concentrations is prepared in triplicate and injected in ability – The data obtained from the injections used for the accuracy run are also usedto calculate ity – The data obtained from the injections used for the accuracy run are also used forthe linearity determination. Two mixtures of the active ingredient and impurities “both" atthe 50 and 150% levels of their target concentrations are prepared in duplicate and evaluatedin the same chromatographic run. If the 50% spiked sample solution contains any impuritiessignificantly above the 0.05% reporting level, then appropriate dilutions of this solutionshould be also be /QL – The 50% spiked sample solution or the “reporting level" solution that was used inthe linearity run may be used to determine the QL of the ucibility – A sample containing representative impurities near their upperspecification limits may be used if one exists. If not, an aliquot of a stock impurity mixsolution should be spiked into a sample. Portions of the same stock should be distributed tothe other laboratories participating in the reproducibility study as appropriate. Six replicatepreparations should be run according to the method ness – Use the sample (spiked with impurities as appropriate) that is used for thereproducibility study under both the standard and modified 30 of 32

Compendium of Analytical ProceduresCAP007/03Drug ProductSpecificity – The active peak of an aged sample is evaluated by UV diode array spectroscopyand an orthogonal technique for peak purity. A placebo and an impurity mix comprised ofall available impurities present at their upper specification level and a sample spiked with theimpurity mix are also run to demonstrate cy (Recovery) – A mixture of the active ingredient, placebo and impurities present at80, 100 and 120% of their target concentrations is prepared in triplicate and injected ability – The data obtained from the injections used for the accuracy run are also usedto calculate ity – The data obtained from the injections used for the accuracy run are also used forthe linearity. Also prepare in triplicate and run a mixture of the active ingredient, placeboand impurities at the 50 and 150% levels of their target concentrations are prepared induplicate and evaluated in the same chromatographic run. If the 50% spiked sample solutioncontains any impurities significantly above the 0.05% reporting level, then appropriatedilutions of this solution should be also be /QL – The 50% mixture or the “reporting level" solution that was used in the linearity runmay be used to determine the QL of the ucibility – A sample containing representative impurities near their upperspecification limits may be used if one exists. If not, an aliquot of a stock impurity mixsolution should be spiked into a sample. Portions of the same stock should be distributed tothe other laboratories participating in the reproducibility study as appropriate. Six replicatepreparations should be run according to the method ness – Use the sample (spiked with impurities as appropriate) that is used for thereproducibility study under the standard and modified 31 of 32

Compendium of Analytical ProceduresCAP007/03AMENDMENT HISTORYDateAuthorised1 March 19958 August 2000AmendmentIssue0102Initial IssueReason for IssueCurrent regulatory expectations, e.g., ICH are now met and thesequence and nomenclature of the validation parameters betteralign with the ICH guidelines. Information on the One-Teamapproach and revalidation of existing methods issue of GSK r 200203Page 32 of 32