Validation of analytical methods

Adrian Covaci

Toxicological Center, University of Antwerp

Why validate?

Need to ensure that quality of analyses is good and stays good

= consistent, reliable and accurate data

Need to ensure a way of standardization between methods

Validation requested by most regulations and QC measures that impact laboratories

The objective of validating an analytical method is to demonstrate its

“suitability for its intended purpose”.

When validating?

Analytical methods need to be validated, verified, or revalidated:

• Before initial use in routine testing

• When transferred to another laboratory

• Whenever the conditions or method parameters for which the method has been validated change (e.g., instrument with different characteristics or samples with a different matrix) and the change is outside the original scope of the method.

Method

Validation

Identification

of problem

Quality

Control

Method

Development

Apply method to real samples

Quality control and quality assurance – more details in the presentation of Felix

Various validation schemes

Harmonized guidelines between International Conference on Harmonization (ICH) of technical requirements for registration of pharmaceuticals for human use (ICH Q2(R1), AOAC International (AOAC), and the IUPAC - for single laboratory method validation.

US FDA - guidance document for analytical method validation for bioanalytical methods.

European Medicines Agency (EMA, formerly EMEA) - “Guideline on bioanalytical method validation”.

Which guidelines to follow for validation?

Eurachem guide “The Fitness for Purpose of Analytical Methods” and NordVal “Guide in Validation of Alternative Proprietary Chemical Methods” - not limited to a specific technique.

More specialized guidelines, such as validation guidelines for pesticide residue analysis in food and feed by SANCO/12571/2013

Or

European Commission Decision 2002/657/EC implementing the Council Directive 96/23/EC,

- to ensure the quality and comparability of analytical results generated by official control laboratories.

No (official) validation guidelines for wastewater analysis!

Which guidelines to follow for validation?

Which guidelines to follow for validation?

Differences between the various guidelines:

- In the requirements (nr days/replicates for a parameter)

- In the established limits (no limits in ICH gudelines)

Choose wisely according to your purpose!

Definitions

Internal (surrogate) standard:

- added in a constant amount to samples, as early as possible in the sample preparation

- added also in the blank and calibration

- used for calibration by plotting the ratio of the analyte signal to the internal standard signal as a function of the analyte concentration.

- is very similar, but not identical to the analyte(s). This results in similar effects during the sample preparation and analysis.

Recovery standard:

- Added in the final extract before analysis

- Used to calculate recovery of IS in samples

Quantitation methods

- external calibration

- calibration with internal standard

- standard addition

- calibration graph in solvent or in the extract of an appropriate matrix –the so-called matrix-matched calibration (the standard addition methods are intrinsically matrix-matched).

For matrix-matched calibration, a blank matrix sufficiently similar to the sample matrix is needed

Quantitation methods

Using an internal standard (IS) can help correcting for sample preparation and to a lesser extent also for ionization suppression.

Two types of IS can be distinguished:

- structural analogs of the analytes

- isotopically labeled internal standards (ILIS).

ILIS may either be isotopically labeled analyte or some other isotopically labeled compound

Which parameters to validate?

Linearity

LOD and LOQ

Precision

Accuracy

Recovery

Qualitative aspects

Matrix effects

Robustness

Linearity

How many calibration points? How are they distributed over the range?

Tests for linearity

- Plot of residuals (absolute or relative)

- Lack of fit (residual variance)

- Zero testing

- Include or exclude origin?

- Weighting of the curve (non, 1/x, 1/x²)

- Non-linear calibrations (e.g. quadratic or

polynomial)

- Range: LLOQ – ULOQ

Calibrators: in solvent or matrix?

Matrix – blank!! Not for influent/effluent WW

– OECD wastewater? Tap water? Milli Q water? Surface water?

Linearity

Linearity

- What to do when measurement is outside the calibration range?

- Re-analyze the sample by taking less sample and adding more IS

- Dilution will not always help

Sensitivity of the analysis – slope.

Calibration curves on different days (check differences in slopes)

Slope should be similar between different calibration batches (e.g. Between weeks, months).

For ILIS, a ~ 1 in y = ax+b

Where y = Ax/AIS and x = Cx/CIS

LOD and LOQ

LOD – defined as the concentration which leads to S/N = 3

LOQ – defined as the concentration which leads to S/N = 10

Instrumental LOQ (pg injected) – a measure of the instrument performance, can be done with standards or extracts.

Method LOQ (ng/L) – a measure of the method performance, done with extracts.

EMA/FDA require <20% bias and <20%imprecision at LLOQ

More information will be given by Tito

Precision

- Describes the agreement between repetitive individual measurements and is calculated as the coefficient of variation (CV) or RSD %.

- Should be demonstrated for the LLOQ, low (3xLLOQ), medium (around 50% of the calibration curve) and high (75% of ULOQ) concentrations, within-day and between-day.

- The imprecision should be <20% for the LLOQ en <15% for the other levels.

Precision

Repeatability (intra-day precision)

- Same operating procedures over a short period of time

- Measurements on the same day (n=5 or 7)

Intermediate precision (inter-day precision)

- Measurements on different/consecutive days (n=5 of 7 per day)

- D = 3 to 5 days

Reproducibility (overall precision)

- Precision between laboratories

Trueness/Accuracy

- Describes the agreement between the measured value and the acccepted reference value and is expressed as percentual difference from this accepted value.

- Should be measured on samples with a known concentration of analyte (QCs).

- Should be measured for LLOQ, low, medium and high concentrations, within-run and between-run (similar to precision).

- Should be <20% for the LLOQ en <15% for the other levels.

Accuracy

- Measurements on the same day (n=5 or 7)

- Measurements on different/consecutive days (n=5 of 7 per day)

D = 3 to 5 days

Spiked samples = QC samples?

SRMs, CRMs – not always available.

Can also be QC samples during analysis of real samples.

Associated with the uncertainty of the method

Recovery

- Calculated as the percentage of analyte signal after sample preparation against the blank matrix in which the same concentration analyte was added.

- Should be evaluated at at least 2 levels: 3 x LLOQ (low) en 75% ULOQ (high).

- For 5 days, the following samples are analyzed: 1 spiked sample and 1 blank sample where the analyte was added after extraction. The IS is always added after the extraction.

Recovery

Recovery of analytes

- Done using spiked (QC) samples

Recovery of internal standards

- Spiked in each sample or QC

- Should be monitored during analysis of sample batches

- SD is also an important parameter. A method has to have a low SD (be reproducible and stable), even if recovery is lower.

- When using ILIS, recoveries can be also lower, but stable.

Matrix effects vs. Recovery

Matrix effects

- Matrix effects are changes in the signal due to the presence of co-eluting analytes or other interferences in the extract.

- Supression or enhancement of the analyte is dependent on the matrix, the quality of sample preparation, the additives in the mobile phase and the ionisation type.

- Should be assessed for different batches of matrices (n=6?, RSD<15%)

- To be assessed for each analyte and internal standard (see next slide)

- Absolute matrix factor calculated as on the next slide

- Relative matrix factor = ratio between absolute matrix factor of the analyte and the corresponding IS.

- For ILIS, relative matrix factor ~1

Matrix effects vs. Recovery

Robustness (or ruggedness)

Ability of an analytical method to remain unaffected by

- small variations in method parameters (mobile phase composition, column age, column temperature, etc.)

- influential environmental factors (room temperature, air humidity, etc.)

It characterizes its reliability during normal usage.

Considering robustness as a method development parameter, EMA and FDA guidelines are missing the term.

Robustness (or ruggedness)

Recommended method parameters to be investigated during robustness studies

Selectivity

- Rather good to very good for MS (or HRMS) methods

Selectivity is the capability of a method to measure ondubbelzinnig the analyte(s) and to differentiate the analyte(s) and the internal standard(s) from components possibly present in the samples.

Such components are typically endogeneous matrix components, degradation products, impurities, etc, but can be also other analytes!

More information in the presentations of Tania (related to validation of qualitative methods) and Robert (related to identification and confirmation of analytes).

Stability

- Calibrators (in organic solvent)

- Calibrators (in water-organic solvent)

- Stability throughout the sample preparation

- Extracts in organic solvent

- Extracts in initial mobile phase

- Different temperatures (+4°C, +20°C - autosampler)

Freeze-thaw cycles

Short-term temperature stability

Long-term stability

Stability

What to do when you want to add a new compound to a validated method?

Partial validation, which parameters to be changed?

All parameters for this compound!

- Accuracy

- Precision

- LOQ

- Linearity

Food for thought

No official guidelines for WW

Which guidelines to follow?

Should we propose guidelines for method validation for ww analysis?

Suggested literature

EMA- ICH- FDA - guidelines

Matrix effects: Matusewzki et al., AC 2003