Sample extraction: recovery, matrix effects and reproducibility

A clear extract and a stable signal do not demonstrate that preparation has recovered the fraction of interest. Analyte may remain in the matrix, adhere to surfaces or change chemically; other substances can alter its instrumental response. Assessing extraction requires a defined analytical question and comparisons that distinguish these possibilities.

Define the analyte, matrix and fraction to be measured

Describe sample condition, heterogeneity and the chemical form to be determined. Measuring total content, a fraction extractable under specified conditions or a free species represents different objectives. A more vigorous treatment can change the very fraction being measured. The result must therefore refer to a defined measurand, rather than generally to “everything the instrument detects”.

Connect sampling, storage, test portion and preparation. Efficient extraction of an unrepresentative aliquot cannot correct sampling. The ICH Q14 framework for analytical procedure development helps connect the purpose with preparation variables. First define the performance required of the result and the matrices, concentrations and conditions within intended use.

Select the approach for the separation required

Dissolution may make an analyte accessible without separating it from other components. Solid–liquid extraction depends on release from the matrix; liquid–liquid extraction depends on distribution between phases and their separation. Solid-phase extraction requires consideration of retention, washing and elution. Precipitation may simplify the matrix while also removing analyte. These descriptions guide selection; they are not interchangeable recipes.

Compare selectivity, compatibility with the subsequent measurement, stability, transfer count, available quantity and safe material handling. A cleaner extract may come at the cost of recovery; greater recovery may bring more interference. Demonstrate the choice for the complete system, including vessels and volumes, rather than inferring suitability from the solvent class or technique name alone.

Identify the variables that require control

Describe extractant composition and quantity, the sample-to-extractant relationship, any relevant pH, addition order, contact time, mixing and thermal conditions. Record separation steps, transfers and delays before analysis. “Mix well” or “allow to extract” does not support comparison between executions when motion or time influences the result.

During development, assess potentially critical variables and plausible interactions: more solvent can simultaneously change the extraction ratio, concentration and capacity of the following step. In routine use, follow the approved method; do not optimise it informally for one sample. Check dilution factors and final volumes, distinguishing them from the nominal volume added.

Separate recovery, matrix effects and stability

Recovery needs an operational definition: which quantity or response is compared, against which reference and at what point in preparation? A reduced response may indicate analyte loss or response suppression. An increased response does not establish that more analyte was extracted.

The original Matuszewski study of HPLC-MS/MS methods distinguishes matrix effects from recovery assessment. This is a specific bioanalytical context: different techniques do not necessarily exhibit the same effects. In LC-MS/MS, comparisons of addition before extraction, addition after extraction and a matrix-free solution can be informative, provided amounts, final volumes and compositions are comparable.

Assess stability separately under the conditions actually experienced. A degradation product, interference or different chemical state can change the signal without representing simple physical loss. An internal standard helps only with effects it has been demonstrated to track; adding it at the end does not automatically document what happened earlier.

Design comparisons around a hypothesis

Use blanks appropriate to the question: reagents, the complete procedure or analyte-free matrix, where available and representative. Account for native content and the volume introduced with the addition. In an LC-MS/MS study, Bienvenu and colleagues show that initial concentration and spike level affect interpretation. Their experimental design does not impose a universal matrix count on every QC method.

The third edition of the Eurachem guide, 2025, cautions that added analyte may be less strongly bound to the matrix than analyte already present. Do not turn a favourable spike result into complete evidence of native-analyte extraction.

Assess precision and robustness as distinct properties

Prepare independent replicates from the start of the step under investigation. Reinjecting the same extract primarily assesses subsequent measurement, not extraction variability. Repeatability concerns short-term, consistent conditions; intermediate precision includes relevant variations within the laboratory. In validation, reproducibility normally concerns between-laboratory comparison: always state which conditions varied.

Robustness explores the effect of deliberate changes in conditions within a justified design. If operator, matrix and equipment change together without an interpretable plan, the effect cannot simply be attributed to one factor. Set criteria and analysis before testing, and include matrices and levels that genuinely challenge intended use.

Document and investigate anomalous recovery

Simulated case. An addition to a suspension before extraction meets its specified criterion, but a real sample gives lower results with the new preparation than with a qualified comparator. Added analyte might be freely available while native analyte remains associated with the matrix. This is a hypothesis: comparator selectivity, homogeneity, stability and calculations also require examination.

Retain data, restrict use of the new configuration and define an investigation using representative materials. Do not apply an invented recovery factor or repeat until the expected value appears. Record identities, quantities, times, conditions, lots and deviations in the traceable preparation workflow. Changes require impact assessment, with verification or revalidation proportionate to risk and approval before use.

EU GMP Chapter 6 provides the framework for QC procedures, records and investigations. Explore related steps in the liquid handling and sample preparation area.

For the related steps, see: Sample concentration and evaporation: losses and recovery; Homogenisers: Selection and Controls in Sample Preparation.

Sources and scope

Consulted on 30 September 2026. ICH Q14, EMA Step 5 revision 1; EU GMP Chapter 6, 2014; Eurachem, Fitness for Purpose, third edition 2025, recovery and precision sections. Original studies: Matuszewski et al., 2003, and Bienvenu et al., 2017, abstracts consulted. No universal extraction conditions or thresholds are inferred from these examples. Table and case are original analytical aids; the case does not describe experimental data.