Two guidelines of the International Conference on Harmonisation (ICH) contain standardised terminology, a verified model of requirements for the validation of analytical procedures, and guidelines concerning the practical aspects of conducting validation studies. Marco Zeller discusses the practical application of these formal requirements for the validation of analytic limit tests specifically for thin layer procedures, which may be used for quantitative and qualitative determinations of impurities in pharmaceutical testing.
Thin-layer chromatography (TLC), is one of the still predominant identity and standard limit tests for impurities, especially for those that are difficult or impossible to detect with liquid chromatography in pharmaceutical industry. The General Chapter on TLC in the European Pharmacopoeia includes quantitative TLC using scanning densitometry and provides the acceptance criteria for resolution and limit of detection. The ICH guideline defines and lists validation characteristics for limit tests and the second gives more details how to perform the experiments.
The process of validating a procedure cannot be separated from its development as the analyst will not know whether the procedure and its performance are acceptable until validation has been performed. However, following best practices requires a highly formalised protocol. Thus, before outlining the validation protocol and the experimental design, it is necessary to qualify all instrumentation and equipment. All chemicals, TLC plates, and reference standards must be defined, specified, and tested. The analytical procedure must be developed, optimised, and documented. A validation protocol, which includes the acceptance criteria and the specified statistical approaches, must be agreed upon and signed.
A quantitative TLC purity test requires a comprehensive validation study - including accuracy, precision, specificity, detection limits, quantitation limits, linearity and range parameters.
TLC is widely used for identification. To demonstrate specificity, the sample, standard substance(s), and pure and spiked placebos are chromatographed simultaneously. Matrix effects must be eliminated to obtain undisturbed spot-shape and Rf-values. In case of an API, the drug substance, its known available impurities, and the API spiked with the impurities at specification level are chromatographied simultaneously. For characterising the specificity, Rf, RS, and A0.05 (RS „ 1; 0.8 ae A0.05 ae 1.2) values are calculated. [Rf (retardation factor) = distance travelled by the centre of the spot/peak ÷ distance travelled by the mobile phase front. RS (peak resolution) = 1.18 x distance between 2 adjaced spots ÷ sum of the 2 peak width at half height. A0.05 (tailing factor) = the width-ratio of the 2 halves of the peak (front/tail) measured at 5 per cent height.]
Specificity of unknown / unavailable degradation products validation of specificity (selectivity) based on forced degradation tests must also be performed, particularly when the analytical procedure is considered to be stability indicating.
Linearity and range
Unfortunately, the term alinearity' is used in both ICH and USP, guidelines although there are numerous analytical techniques that indicate/demonstrate or lead to a nonlinear relationship between the concentration of analytes in samples and the corresponding measurement signals. In addition, a measurement directly proportional to the analyte quantity should not automatically allow a one-point calibration in routine measurement. Therefore, the validation parameters to be determined are not only the mathematical signal to amount correlation in the selected working range, but also the verification of the proposed calibration model in a routine analysis.
In TLC, and especially in the case of evaluation performed by scanning in the UV/VIS reflection mode, most calibration functions are nonlinear. The well-defined concentration ranges and specification limits in pharmaceutical analyses often make it possible to find and use a quasi-linear calibration function where there is no significantly better fit of the calibration data to a nonlinear (eg polynomial) regression. Therefore, for routine analytical procedures, a three-point calibration model must be used, also making it compliant with the defined working range described in the monograph of the European Pharmacopoeia.
For proving linearity or basic calibration of each impurity in the proposed working range, a 10-point calibration graph is constructed from not less than eight independent and equidistant concentrations, statistically distributed. The highest and the lowest concentration are spotted twice. Residuals of the regression line are plotted against the applied quantities. The calibration function can be considered linear if the residuals are distributed around the 0-line at random, without any trend. The check for linearity can be done using various statistical approaches. The acceptance criterion for the linear range are the Quantitation Limit up to 120 per cent of the specified limit of the impurity.
Precision provides information about the random error of the analytical procedures including sampling and sample work-up. Precision is expressed as a relative standard deviation (RSD) of multiple determinations of a homogeneous sample.
The same real sample is determined on the same plate by applying at least six analytical solutions from individual weightings. Repeatability provides information about the variation caused by sample preparation, sample application, and evaluation within one analytical run and within a short period of time. The same sample is determined on the same plate by applying at least six analytical solutions from individual weightings.
Intermediate precision
The ICH guidelines give no guidance on how to determine intermediate precision. Two approaches are most common.
a) Repeatability test runs are performed on different days with a different combination of analyst and equipment but in the same laboratory. Mean and RSD values of both sets of data are then compared. b) At least 6 replicated determinations are performed on different days with a different combination of analyst and equipment, but in the same laboratory. The RSD value for this set of data is then calculated. Table 3 shows acceptance criteria.
Reproducibility describes the analytical variability between different laboratories at different sites. It is determined by interlaboratory (round robin) tests and is not a compulsory part of the analytical validation in pharmaceutical analysis.
Accuracy expresses the closeness of agreement between the value that is accepted as a convential true value or an accepted reference value and the result found.
For determination of possible bias, triplicate analysis of solutions of the analyte spiked with three different concentrations of the available impurities is performed. The recovery is calculated as a percentage value of the added amount of impurity (Table 4).
For finished product testing, determination of accuracy is even more important because it verifies the quality of the delicate sample preparation step. Acceptance criteria for impurity determinations in finished products are close to approximate those given in Table 3 for APIs. Wider ranges may be acceptable in certain cases.
Alternatively, accuracy may be proven by constructing recovery function comprised of at least 6 concentrations covering the working range of impurities, ranging from the QL up to 120 per cent of specified level. Found concentrations are plotted on the ordinate versus the original calibration concentrations and the abscissa, a recovery curve is obtained. The recovery function can be accepted if the given confidence interval of the intercept includes the origin and that of the slope includes 1 (generally, x = 0.05, p = 95 per cent).
Although the determination of the detection limits (DL) is not compulsory in the validation of a quantitative impurity test, knowing this data can be useful in establishing the impurity profiles of substances. For determination of the DL and quantitation limits (QL), solutions of the expected impurities and the main compound of the matrix (for finished products n „ 3) are applied in decreasing quantities, in triplicate. The same volume of the pure solvent as a blank is also applied. A calibration graph is constructed by plotting the peak heights or areas against the applied quantities of the substances investigated. The DL and QL are calculated based on the signal-to-noise ratio (see Table 5 and 6).
Conclusion
Contemporary TLC is a chromatographic technique that fully complies with cGMPs. The results can be objectively documented in computerised form and on hard copies qualitative and semi-quantitative testing as a video picture or video densitogram and quantitative assays or tests as a densitogram.
Enter 45 or at www.scientistlive.com/elab
Marco Zeller is Product Manager Analytic, Solvias AG, Basel, Switzerland. www.solvias.com
