ANALYTICAL DEVELOPMENT Assay

HHandbook of P GGeneric DDevelopment

ASSAY VALIDATION

Ruggednessand

Robustness

uggedness and robustness testing isapplicable to pharmaceutical R&Dand QC Laboratories, especially

when methods are transferred anddistinguishes the similarities and differencesbetween the two analytical concepts andhow they are applied in the laboratory.

Analytical methods are usually developed inan R&D laboratory and eventuallytransferred to the production quality controllaboratory for routine product analysis. Thisprocess of technical data transfer from onelab to the other requires a cleardemonstration that the methodology can besuccessfully transferred. How is it done?

There is an easy-to-use procedure to meetthis regulatory and compendial requirement.

Ruggednessis a USP

RequirementRobustness

is not.Ruggedness and Robustness.The USP defines ruggedness as "the degreeof reproducibility of test results obtained bythe analysis of the same samples under avariety of normal test conditions such as:

♦ Different laboratories

♦ Different analysts

♦ Different instruments

♦ Different reagent lots

♦ Different analysis days

♦ Different elapsed assay times

♦ Different assay temperatures …"

These factors are all external to the writtenanalytical method and each parameter shouldshow a lack or indeed absence of influenceon the test results obtained.

But what about the internal factors of thewritten test method such as a change in theflow rate (mL/min) or the concentration ofthe organic acid in mobile phase (HPLCsystems) or better still, a change from aPhenomenex Bondclone™ 10µ C-18column to a Waters µ-Bondapak™ 10µ C-18 column? These small but deliberateinternal variations in method parameters ofthe written analytical procedure should beevaluated to access whether the analyticalprocedure remains unaffected by these slightchanges.

Robustness is defined by both the USP andthe ICH Tripartite guidelines as "a measureof its capacity to remain unaffected by smallbut deliberate variations in methodparameters and provides an indication of itsreliability during normal use " Robustnessis defined both in the USP and ICH, but isnot required.

Furthermore, ruggedness measures the lackof external influence on the test resultswhereas robustness measures the lack ofinternal influences on the test results.Internal and external variations are oftenmixed together in analytical validationpackages erroneously, under the heading ofone or the other. Internal and externalanalytical parameters should be separated

R

ANALYTICAL DEVELOPMENT

HHandbook of -GGeneric DDevelopment

and appraised individually. A simpleexperimental design can evaluate bothruggedness and robustness, as separatedistinguishable entities - together.

Table 1 Comparison Table.

Attribute. Ruggedness. Robustness.

USP ValidationRequirement

þý

ICH Validation

Requirement ý ý

Internal change ý þExternal change þ -

MethodVariations

- þ

Environmental

Variationsþ -

Designing Analytical ExperimentsSuch a designed experiment candemonstrate that methodology andenvironmental factors may or may notinfluence the test results. It is hoped that theanalytical method is both rugged and robust,however a well designed experiment mayidentify test conditions or specification limitsthat need to be closely controlled andtightened or even test parameters that needfurther investigation and optimization.

Advantages - Designing the ExperimentA designed experiment is a simple matrixdesign. The Plackett-Burman designs aremost applicable to technical transfer of avalidated analytical methods fromdevelopment to quality control centres. Forthe most cost-effective design, the attributesof both R&D and QC laboratories should beincorporated into development validationprotocol of the assay method.

The advantages of these designs are quitesimple - the number of tests required issimply dramatically reduced. 56 assays (i.e.7 x 8) are needed to evaluate seven internaland/or external variables, these can bereduced to eight quick assays using an eight-run Plackett-Burman design.

Even more dramatic, for eleven variables(11 x 12) a minimum of 132 one-factor-at-a-time data points would be required, but viamatrix testing using an twelve-run design,only 12 HPLC assays are needed to producethe equivalent of 132 individual one-factor-at-time assays.

Few, if any HPLC assay analytical methodscould have more than 12-15 significantenvironmental or method variations.

Table 1.

An Eight Run Design Template

TestExternal / Internal

Changes / VariationsASSAY

TESTNo A B C D E F G RESULT

1 + + + − + − − 99.3

2 − + + + − + − 101.5

3 − − + + + − + 100.4

4 + − − + + + − 97.9

5 − + − − + + + 98.5

6 + + − − + + 99.0

7 + + + − − + 97.9

8 − − − − − − − 100.9The (+) or (-) signs are used as variables in the 8 rundesign. Assign (-) to Analyst I ; Day I; Column I and(+) to Analyst II; Day II; Column II and so onA to G are chosen as the external variations(ruggedness) anticipated to arise during use in theDevelopment Lab.

Table 2. 8 RUN DESIGN

Template for Ranked Effect and Means

External & Internalchanges/variations

RankedEffects

M

valuesA - Analyst I & II 1.8 -1.35

F - Analyst III & IV -0.76

G - Reagents I & II -0.35

E - Week I & II 0

B - Week III & IV +0.35

C - Column I & II +0.76

D - HPLC No I & No II 0 +1.35

The M values are obtained from statistical designtables. The ranked Effects are calculated by simpleaddition of assay test results and then dividing byhalf the number of runs (i.e. 4 in a 8 run design).

ANALYTICAL DEVELOPMENT

HHandbook DDevelopment

Calculating the Ranked EffectsFor A Figure 1.

99.3 +101.5 −100.4 −97.9 +98.5 −99.0 +97.9 +

100.9 −-7.2 ∑-1.8 /4

For D Figure 2.99.3 −

101.5 +100.4 +97.9 +98.5 −99.0 −97.9 +

100.9 −0 ∑0 /4

Results.A linear-linear scale is used. Plotting theRanked Effect on the X-axis vs. the Mvalues on the Y-axis produce a normalprobability plot of effects. If a value liesoutside this straight line one can concludethat the method is not rugged / or robust, asclassified, for that particular variable (e.g.[say] flow rate).

12 Run Designs.A template for 12 run design is used (Tables3 & 4), when more than seven factors arepresent. This design will give 11 factors foranalysis. The M values are constant for anygiven design and are actually the means ofthe order statistics (3) for a sample size ofeleven. As they always remain the same, thetemplate can be used for any ruggedness /Robustness validation method protocol. Usea Eight Run for evaluating say, ruggednessonly, and a Twelve Run design for both

ruggedness and robustness when transferringa method to another laboratory.

Table 3.

An Twelve Run Design TemplateExternal / Internal

Changes / Variations (11)ASSAY

TESTNo A B C D E F G H I J K

RESULT

1 + + − + + + − − − + − 99.32 − + + − + + + − − − + 101.53 + − + + − + + + − − − 100.44 − + − + + − + + + − − 97.95 − − + − + + − + + + − 98.56 − − − + − + + − + + + 99.07 + − − − + − + + − + + 98.88 + + − − − + − + + − + 99.99 + + + − − − + − + + − 100.610 − + + + − − − + − + + 98.911 + − + + + − − − + − + 97.912 − − − − − − − − − − − 100.9

The (+) or (-) signs are used as variables in the 12 rundesign. Assign (-) to Analyst I ; Day I; Column I and (+) toAnalyst II; Day II; Column II and so on…A to K are chosen as the external (ruggedness) / internal(robustness) variations anticipated during the transfer fromthe R&D to QC laboratory.The assay results are entered into the table on completion ofthe12 HPLC assay analyses.

Table 4.Template for Ranked Effect and MeansExternal & Internalchanges/variations

RankedEffects

Mvalues

A- R&D & QC Lab -1.59F- Day I & II -1.06G- Analyst I & II -0.73E- Analyst III & IV -0.46B- Reagents I & II -0.22H- [Solvent] I & II +0.00C- Heating Rate I & II +0.22J - Column I & II +0.46K- Temperature I & II +0.73I - Flow rate I & II +1.06D- Elapsed time I & II +1.59

The M values are obtained from statistical design tables.The Ranked Effects are calculated by simple addition of(-) (+) assay test results and then dividing by half thenumber of runs (i.e. divide by 6 in a 12 run design).

The effects form A - K were selected as the mostsignificant variables between the two labs. Templates areavailable for up to 100 run variables (100 x 99).

Sum the Assayvalues byassigning

a positive ornegative valueobtained from

the 8 rundesign

Perform thisaddition for

each of the eightvariables and

divide thesum by 4

(half the numberof runs)

ORAL TABLETS ANALYTICAL DEVELOPMENT CHAPTER 13

HHandbook of PPharmaceutical Sect: 13.13.4141 GGeneric DDevelopment

Method Procedure.

1. Choose the number of variables required

and select a run design template.

2. Assigning the minus (-) or plus (+) values:These are arbitrary designations. As astandard rule assign a 'minus' (-) to I or alower limit and a 'plus' (+) to II or a upperlimit. Evaluate a range limit by assign (-)value for lower and (+) value for higher (i.e.Flow rate 1.2 mL/min assign (-) and 1.8mL/min assign (+)). Likewise Day I assign (-) and Day II assign (+) and so on…

3. Perform the HPLC assays in a randomorder.

4. Tabulate the assay results in the template.

5. Calculate the Effects (Figures 1 and 2).

6. Rank the Effects from smallest to largest.

7. Plot the Effects against the M values.

8. Evaluate the plot.

Conclusion.The results from the plot form a nearstraight line. It can be concluded that theanalytical method is (a) rugged for theexternal factors over the tested range and(b) robust for the internal factors over thetested range in the 12 run design.

Figure 3.

A Normal Probability Plot of Effects

+1.5*

M*

VA *LU *ES *

*

*-1.5

-8 Ranked Effects +7

Process Qualification Stage.The evaluation of ruggedness androbustness should be finalised at the endof the development phase - around thetime of the process qualification lotmanufacture. Theruggedness/robustness evaluation shouldbe developed with the commerciallaboratory equipment in mind. It shouldshow the reliability of an analysis withrespect to deliberate variations in themethod parameters.

Ruggedness/robustness determinationsare essential when transferring analyticalmethods from the developmentlaboratory to the commercial plantquality control laboratory. There mayusually be a difference in columns orHPLC machine models used.

A consequence of ruggedness /robustness evaluation is that a series ofsystem suitability parameters areestablished to ensure that the validity ofthe analytical procedure is maintainedwhenever used.

References:

1. "Validation of compendial methods" USP 23<1225> USPC Rockville Maryland USA 1994.

2. International Conference on Harmonization"Guidelines on validation of AnalyticalProcedures: Definitions and Terminology…;Federal Register (March 1, 1995.)

3. "Validation of compendial methods" USP 23<1225> USPC Rockville Maryland USA 1994.

4. USP/NF XXIII USPC Rockville Maryland USA1994.

5. Scale up and Post approval ChangesManufacturing and Controls In vitro Dissolutionand In Vivo Bioequivalence DocumentationCEDER 1995 (SUPAC)

6. ASTM Standard Guide For ConductingRuggedness Tests E1169 American Society fortesting Materials Philadelphia 1989.

7. Kateman and L. Buydens, The Ruggedness TestQuality Control in the Analytical chemistry JohnWiley and Sons NY 2nd Edition 1993, pp118125.