Aspects of Bioanalytical Method Validation for the Quantitative Determination of Trace Elements

Abstract

Bioanalytical methods are used to quantitatively determine the concentration of drugs, biotransformation products or other specified substances in biological matrices and are often used to provide critical data to pharmacokinetic or bioequivalence studies in support of regulatory submissions. In order to ensure that bioanalytical methods are capable of generating reliable, reproducible data that meet or exceed current regulatory guidance, they are subjected to a rigorous method validation process. At present, regulatory guidance does not necessarily account for nuances specific to trace element determinations. This paper is intended to provide the reader with guidance related to trace element bioanalytical method validation from the authors’ perspective for two prevalent and powerful instrumental techniques: inductively coupled plasma-optical emission spectrometry and inductively coupled plasma-MS.

Bioanalytical methods are used to quantitatively determine the concentration of drugs, biotransformation products or other specified substances in biological matrices, such as blood, tissue, serum, urine and plasma. These methods often provide critical data to nutritional, safety, pharmacokinetic/pharmacodynamic or bioequivalence studies in support of regulatory submissions. As a result, it is important to ensure that a bioanalytical method is capable of generating reliable, reproducible data. To that end, the bioanalytical method validation process is designed to unequivocally demonstrate that a bioanalytical method is appropriate for its intended use, including the important concepts of reproducibility and transportability. In the context of this paper, method validation encompasses an established bioanalytical method having undergone method development to demonstrate and confirm that it is fit for the intended purpose. Robust method validation is required to satisfactorily yield reliable, reproducible results that meet or exceed performance requirements recommended within the applicable Regulatory Guidance for the fundamental parameters of accuracy, precision, reproducibility, selectivity, sensitivity and stability [101–104].

The marketplace prevalence of small, organic-molecule therapeutic agents and the increasing importance of macromolecular therapies have focused the attention of industrial committees and regulatory agencies towards bioanalytical method validation for these species [1]. For small-molecule measurements, a chromatographic technique is commonly coupled to a mass spectrometric detector (e.g., LC–MS/MS), while macromolecular measurements are typically conducted with non-chromatographic ligand-binding assays (LBAs). As a result, best practices for bioanalytical method validation of small and macromolecular compounds contain characteristics related to the size, structure and detection modality of these species [2–10]. In contrast, there is currently limited information available related to bioanalytical method validation for the determination of trace elements. Furthermore, this limited information is not always standardized across the bioanalytical support industry.

Trace elements are important components of many biological systems and therapeutic agents, and subtle concentration changes within an organism can have a significant impact on nutrition and the genesis, prevention and treatment of disease. As a result, their accurate, reproducible determination in biological matrices is often critical to the ultimate success of pharmacokinetic, bioequivalence, toxicological and epidemiological investigations. Although the validation approach for trace elements is similar to that utilized for small-molecule organic compounds, there are some important nuances for trace element bioanalytical method validations that stem from their physiochemical properties. For example, trace elements are ubiquitous in the environment, so attention to controlling the analyte contribution from the reagents and the analytical procedure is an absolute necessity. Trace elements are also often endogenous to biological matrices, and normal levels within a matrix can vary significantly between samples as a function of diet and other environmental exposure or lifestyle behaviors. As a result, emphasis on achieving adequate homogenization and digestion of a suitable control matrix is critical, even for non-endogenous elements. Furthermore, the total concentration of a trace element is generally stable, while the stability for different species of the same element (chemical form, including metabolites, or oxidation state) can vary significantly. As a consequence, attention to potentially different volatility, precipitation, and container wall adsorption characteristics for the expected chemical form or species of a trace element should be noted during method development. For total trace element determinations, an aggressive digestion approach is typically appropriate, whereas a milder digestion or an extraction procedure is often warranted when it is important to maintain species integrity or separation of the analyte from a difficult matrix is possible.

Although there are many analytical techniques that can be used for the determination of trace elements in biological matrices, the current prevalence and likely future utility of inductively coupled plasma-optical emission spectrometry (ICP-OES) and inductively coupled plasma-MS (ICP-MS) make these detection modalities the focus of this paper. These powerful techniques provide a high level of sensitivity, typically low parts-per-billion (ICP-OES) and low parts-per-trillion (ICP-MS) quantitation limits, while offering multi-elemental analysis capability. In addition, they allow for rapid, accurate and precise elemental determinations that are relatively free of interferences. It is important to note, however, that the following discussion would also apply to other atomic spectrometric techniques, such as atomic absorption, atomic fluorescence and flame atomic emission.

This paper is intended to provide the reader with guidance related to trace element bioanalytical method validation. It is also intended to convey the authors’ perspective on the application of the current Bioanalytical Method Validation Guidance [101–104] and the 3rd AAPS/US FDA Crystal City Bioanalytical Workshop white paper [3] to trace element analysis. Although the validation principles described within the Guidance, and the principles of GLP can be applied to bioanalytical trace element analysis, there is currently no documented guidance or white paper available for this application, specifically. Furthermore, considerations of current efforts to harmonize global bioanalytical method validation guidance, and industry and regulatory consensus [11–16] are provided within. Ultimately, it would be desirable for trace element analysis methodology to be included in subsequent harmonized Bioanalytical Method Validation Guidance.

Validation parameters

Bioanalytical Method Validation Guidance addresses the core validation parameters of accuracy, precision, selectivity, sensitivity, reproducibility and stability. The relevance of white papers has become more important recently since the FDA is in the process of revising its current guidance, and the European Medicines Agency is revising its 2009 draft guidance. Furthermore, subsequent consensus meetings have identified many outstanding and new aspects of method validation that are critical to ensure method robustness, data integrity and regulatory compliance [11,17–19]. During validation, data for core validation parameters and a suite of supporting parameters are identified, and method performance is evaluated. Fundamental and supporting validation parameters are discussed in detail below and are summarized in Table 1.

Although many ICP-OES and ICP-MS instrumental analysis parameters that influence system performance are considered tunable, and can be optimized to achieve the best possible instrument performance on each analysis day, other parameters should be set in the method validation and maintained definitively throughout sample analysis. Typical nontunable parameters for ICP-OES and ICP-MS systems include sample pump tubing size and uptake rate, spray chamber type and temperature, nebulizer type, analyte and IS emission wavelengths or isotopes, and interface cones (for ICP-MS), while tunable parameters can include nebulizer flow/pressure, torch position and ion optic lens voltages (for ICP-MS). It is critical to categorize and define tunable and non-tunable parameters prior to the onset of formal validation testing. For a summary of general tunable and non-tunable ICP-OES and ICP-MS instrumental parameters, please refer to Table 2.

Method development

Prior to beginning formal validation, it is prudent for the analytical laboratory to thoroughly test the proposed bioanalytical method during a rigorous method development process. For trace element determinations, method development activities typically include the judicious selection and evaluation of instrumental parameters associated with the analyte and the IS, assessment of analyte selectivity, linearity, sensitivity (analyte quantitation limits), recovery in the presence of control matrix and likely interfering species, potential carryover and sample homogeneity/representativeness. Ideally, the developed sample preparation and analysis procedures would enable consistent analyte determination at biologically relevant levels, while considering potential challenges that might arise during bioanalytical sample analysis (e.g., endogenous analyte concentrations or occasional lipemic samples). If transfer of the validated method is anticipated, consideration should also be given to selection of transportable sample preparation and analysis parameters during method development.

Depending on the application, it may be desirable for the analytical laboratory to conduct speciation experiments to ensure that the analyte can be extracted from control matrix when it is present in a particular chemical form or oxidation state. Speciation experiments are typically conducted using a hyphenated technique, such as ion chromatography (IC)–ICP-MS, or through the use of isotopically enriched, species-specific forms of an analyte. These experiments can often provide useful information and facilitate method development, even if the ultimate objective of the bioanalytical sample analysis is determination of the total concentration of a trace element.

Accuracy & precision

Method accuracy and precision may be regarded as the primary indication of method effectiveness as they describe the extent of systematic and random errors (mean bias and variation, respectively) associated with repeated determinations of spiked control samples (standards and quality control samples). In a full validation, at least five replicates of quality control (QC) samples at each level are analyzed within three independent accuracy and precisions runs, and over at least two different days by two independent analysts for respective robustness [2,101]. In practice, more facilities are analyzing six replicates at each QC level, but whether five or six, it is important that the same number are used when qualifying new levels or preparations of QCs, including dilution QCs. In addition, the qualification of the LLOQ and ULOQ samples are included in accuracy and precision testing to verify the calibration range and limits. In practice, the LLOQ-QCs are analyzed three times during the three accuracy and precision runs. Although use of run qualification QCs in other method validation experiments is necessary to qualify the respective run, the use of frozen QCs must not be used until the applicable frozen stability has been validated.

The acceptable levels of QCs within a calibration range have been well established [2,3,6,7,101]. There must be at least three levels of QCs that are qualified and used within each batch of a sample run. The low-QC is near the LLOQ and is usually three-times the level of the LLOQ, or at least not exceeding this. The mid-QC must be within the middle of the calibration range and ideally at the geometric mean of the low- and high-QC levels. The high-QC must be qualified within the highest quartile of the calibration range. Regulatory agencies have indicated the expectation that at least two QC levels lie within typical data samplings for the applicable study. For instance, as preclinical analysis progresses and an acceptable dose is better defined, it is often observed that analyte concentration data are not as variable and fall within the lower end of the curve. In this situation, there is likelihood that only the low-QC is falling within the sampled data when concentrations are back calculated. This is not acceptable to the regulatory agencies, and it would become necessary to qualify an additional QC level so as to meet their expectation. An alternate approach is to validate a calibration range that is smaller, but QCs will still need to be qualified to meet the acceptable levels described above.

The accuracy and precision requirements for elemental analyses are the same as those used for LC–MS/MS analyses. For a method to be acceptable, it is expected that both the inter- and intra-batch precision (%CV) be <15%, but <20% at the LLOQ level, and the accuracy, usually expressed as absolute mean bias (%RE) be ±15%, and ±20 at the LLOQ level. For accuracy and precision runs, all of the QC replicates are included in the respective calculations. This is also true for the qualification of new lots of QC preparations or additional QC levels intended to be partially validated. Inter-lot accuracy of the LLOQ and ULOQ levels in at least six individual lots of matrices is customarily investigated, although only the LLOQ inter-lot accuracy is currently required. The value of investigating both the lower and upper standard curve levels is to provide assurance that there are no interferences or matrix effects at the extremes of what is typically an unweighted linear standard curve when conducting elemental analysis.

As with other modes of instrumental analysis, ICP-OES and ICP-MS bioanalytical measurements employ QCs and calibration standards that are prepared by fortifying the control matrix with the analyte or analytes of interest. To prepare applicable stock solutions, a well-characterized standard reference material of known identity and purity is required [5]. It is common practice for the elemental analytical chemist to obtain standard reference materials from a certifying agency, such as the National Institute of Standards and Technology (NIST), or stock solutions directly traceable to the reference material. Although appropriate characterization of these solutions may be performed by the certifying agency or the commercial stock solution vendor, it is still necessary to verify the quantification of the reference standard stock solution based on a documented procedure (usually described within a standard operating procedure [sop]).

Internal standardization

An IS for trace element analysis usually refers to an additional element that is added in a constant amount to incurred samples, calibration standards and QC blanks and samples prior to sample extraction or analysis. The IS response is incorporated into the calibration process by plotting the ratio of the analyte signal to the IS signal as a function of the analyte concentration. The IS should have ionization or atomization behavior that effectively mimics those of the analyte, and that will allow for the continuous correction of analyte signal during normal instrument operation, and the monitoring of extraction or digestion efficiency of each sample, standard and QC.

For trace element determinations, the timing of IS addition (before or after digestion) is typically a function of the extent of sample preparation. For small-molecule organic compounds, analytes are typically extracted from biological matrix following a homogenization step through techniques including protein precipitation, liquid–liquid extraction or solid-phase extraction. As a result, incorporation of an IS prior to extraction is necessary because of potentially significant variations in extraction efficiency. In contrast, aggressive acid digestion procedures are often used for the determination of total trace element concentrations, which can result in complete or near complete matrix destruction. These procedures usually provide complete availability of the trace element to the atomic spectrometric instrument, so IS addition before digestion can be redundant. In these cases, samples can be charged with IS after digestion, thoroughly mixing before analysis. Any scientific consideration for the extraction, homogenization and/or digestion techniques used in trace element analysis should be addressed by comparing data integrity from charging an IS before and after digestion during method development and confirming the ruggedness of analyte recovery during validation. For the determination of a particular chemical form or species requiring a milder extraction procedure, addition of IS prior to extraction should be considered.

When using an IS to correct for a change in analyte signal during the normal ICP-OES measurement process, it is important to consider if the IS element is compatible with the (acid) matrix from a solubility standpoint and if there are potential spectral interferences upon the IS emission wavelength. The IS must also be suitably contaminant-free, not present in the sample matrix at measurable, endogenous levels, and be added at a concentration sufficient to provide a good signal-to-noise ratio. The IS element should follow the same pattern of ICP intensity change (similar emission or ionization profile) as the analyte. This is where many potential problems occur, because an IS with the same plasma/temperature behavior as the analyte is difficult to find for each analyte while avoiding the other issues listed above.

In addition to the items noted above for ICP-OES, use of an IS element during ICP-MS analyses should also be employed to correct for potential space–charge effects and salt buildup on the sampler cone interface during measurement. Space–charge effects are thought to occur at the region between the skimmer tip and ion optics, which could potentially result in signal suppression at high analyte concentrations as a function of mass. Differences in instrument design affect the ability to predict the conditions under which the effect is minimal. Salt buildup on the ICP-MS sampler cone results from the presence of high matrix solids, which can result in partial or complete clogging of the orifice.

Calibration range & response

Demonstration of detector response as a function of analyte concentration is a critical component of method validation. Although the 2001 Guidance requires at least six non-zero standards [101], many laboratories are using at least eight [2]. In order for a calibration standard to be considered acceptable and used in the regression, the difference between the back-calculated concentration and the nominal concentration must be within ±15%, except at the LLOQ level where the difference must be within ±20%. If a standard fails to meet the acceptance criterion, it may be excluded from the regression as long as at least 75% of the total standards are acceptable. Currently, atomic spectrometric instrument software limitations normally limit the analyst to one calibration curve unless a laboratory processes data offline (e.g., through use of a LIMS system). In this latter case, it is recommended that all of the samples are bracketed between the first and second standard curves. The advantage to this practice is that any instrument response variations, including a loss of response over the course of a run, can be readily identified by curve divergence or other impacting variations in standard curve values. Regardless of the calibration strategy, the 75% acceptance rule would apply standard curves samples. Although it has been a practice with environmental ICP-OES and ICP-MS elemental analysis to use a zero sample (blank with IS) within the regression analysis, or force the regression through zero, it is usually not acceptable for bioanalysis since quantitation cannot be obtained below the lowest level of quantitation (LLOQ standard). Moreover, forcing a curve through a zero standard or use of anchor points is not necessary since the sensitivity and stability of ICP-OES and ICP-MS instruments is such that calibration curves are predominantly linear and do not requiring weighting.

The order of rejection of calibration standards must be documented a priori and validated, where the acceptable practice is to reject standard levels in the order of those back-calculated concentrations starting at the greatest percent difference from their respective nominal level. The use of outlier tests must be described with rationale prior to the initiation of the validation since this is also validated. If the method is being validated to accommodate curve truncation, then this must be defined a priori, and the standard and QC concentrations should be considered to ensure that at least three levels of QCs still fall within the calibration range if either the LLOQ or ULOQ standard is rejected. It is therefore advisable to validate a calibration curve having two calibration standard levels below and above the LLOQ and ULOQ standards, respectively. It cannot be stressed enough that when the LLOQ standard level is rejected, the next standard becomes the lowest level of quantitation, and therefore any QC level that is below the new LLOQ standard cannot be quantified.

If a suitable, analyte-free matrix is not possible or realistic, and matrix effects are negligible, it should be possible to construct parallel standard curves in the sample matrix and the ‘neat’ digestion matrix. Parallel curves enable the determination of the analyte in the biological matrix against a standard curve prepared in the digestion matrix. A parallel linear regression prepared in the neat digestion matrix would exhibit a similar instrument response (slope) to a regression prepared in the presence of matrix across the entire calibration range, but would exhibit a smaller y-axis intercept due to endogenous analyte levels in the matrix.

Additional controls in elemental analysis

Elemental analysis laboratories have often employed continuing calibration verification (CCV) controls as an analysis QC measure. These controls are often used to assess potential variations in analyte and IS instrument response attributable to some models of ICP instrumentation when ICP-OES or ICP-MS instrument software limitations prevent use of bracketing calibration curves. CCV controls are typically mid-level calibration standards, often without matrix, that are periodically analyzed at predetermined intervals to confirm that the calibration slope throughout an analysis is not impacted by potential instrument response variations. If used, CCV check pass/fail criteria based on instrument response (or determined concentration) are clearly defined a priori. Furthermore, these criteria are validated, and become part of the acceptance criteria for the validated method. In practice, CCV controls bracket samples, and failure of a CCV control results in rejection of bracketed samples. It is critical to note that optional CCV controls are only prepared at one level and do not supersede the use of required matrix QC samples at the low, mid and high levels.

The use of matrix blanks is similar in elemental analysis to LC–MS/MS analysis. Matrix blanks without IS are used to assess interference (blank samples), and control matrix blanks with IS are typically used to assess carryover (zero samples). Matrix blank controls are acceptable when the response is <20% of the LLOQ response. Furthermore, if it is determined during method development that the acceptance criteria for the carryover blank must be higher, for instance when there is an observed endogenous level of the element of choice or homogeneity issues, then the rationale for the adopted acceptance criteria must be described, and it is necessary to document this within the validation report.

Carryover is usually assessed throughout the method validation. It is recommended that at least two carryover blanks are inserted after the highest standard level(s) and the last high-QC grouping. This practice is considered prudent as it allows for carryover troubleshooting and determination of extent when identified (for instance, if there is carryover to the next blank).

Stability

Analyte stability in a biological matrix is a function of storage condition, the matrix, the analyte itself and the sample container system [5,101]. Although the total concentration of a trace element in a biological matrix is generally stable, it is important to note that different chemical forms or species of an element can have significantly different stability profiles. For example, various forms of the same element in a biological matrix, or in a matrix extract after sample preparation, can exhibit vastly different volatility, precipitation and container-wall adsorption characteristics. As a result, it is recommended that the likely form (or forms) of a trace element be considered when assessing stability.

Matrix stability experiments should be conducted using a representative matrix, including the type of anticoagulant for blood-based samples. This is especially critical if the analyte is a nominal component of the anticoagulant or if the anticoagulant otherwise contributes to background from its ability to effectively complex an analyte. Similarly, if a stabilizer is planned for incurred samples, as is often the case with urine, then the same stabilizer should be used with the stability samples (QCs). When matrices are altered, such as being stripped, QC samples must be validated similarly and used as the stability samples even when the calibration curve samples are validated based on an alternate or surrogate matrix. The use of an alternative or surrogate matrix is not recommended, but may be necessary when no alternate strategy is available for the quantitation of elements in the presence of endogenous material, as determined during method development. Regardless of the source of matrix, matrix alterations or interferences, QC samples must be prepared using the same matrix as the incurred samples. This practice is critical to method validation since QCs are representative of incurred samples.

Several types of stability are assessed during method validation. To the extent possible, stability experiments should capture storage conditions that are likely to be encountered during sample collection, storage and bioanalytical sample analysis. The necessary trace element stability experiments are described in detail below.

Stock solution stability

In elemental analysis, it is common to obtain standard reference materials from a certifying agency, such as NIST, or stock solutions directly traceable to the reference material. Although appropriate characterization of these solutions may be performed by the certifying agency or the commercial stock solution vendor, it is still necessary to verify the quantification of the reference standard stock solution based on a documented procedure (usually described within an SOP). Direct use of these standards as purchased prior to expiration can eliminate the need to demonstrate stock solution stability. During method validation, it is important to document the conditions under which the reference and IS, and associated stock reagents, are maintained for stability to meet the basic requirements of the assay performance. When a stock solution exists in a different buffer or organic, the stability of this stock solution should also be assessed. Note that the SOP-defined stability of stock solutions prepared from certified reference materials is valid as long as the stock solution was prepared prior to the expiration of the certified reference material. Should it be necessary to prepare a stock or IS solution from a powder, or if dilution of a traceable, commercially available solution is required, stability should be evaluated under the appropriate storage and relevant sample processing condition (room temperature, refrigerator or freezer) for the useful lifetime of the solution. The stability of stock solutions (prepared either from a certified reference standard or a powder) should be demonstrated at one concentration level in at least duplicate fashion. It is acceptable to use the replicate stock solutions for the preparation of standard and QC samples if the %CV of the replicates be within ±5%, or based on precision in the documented procedure at the applicable laboratory [5].

Long-term, freezer storage stability

Long-term storage stability should be conducted to simulate the storage conditions of incurred samples, including shipping storage conditions. To assess long-term stability, QC samples are prepared for at least two nominal concentrations (ideally, low- and high-QC levels) and analyzed in at least triplicate against a freshly prepared standard curve. For total trace element determinations, storage at -20°C is generally adequate, whereas -70°C storage may be more appropriate for the determination of organometallics or trace element species. It is important to note that the FDA has indicated that samples stored at -70°C should also be assessed for stability at -20°C [3]. It is also recommended that dilution QC stability be assessed (store dilution QCs with the low- and high-QCs), and that sufficient QCs are stored to allow for multiple periods, periods that were unexpected, and for repeat analysis. Since long-term stability extends beyond the period usually required to fully validate a bioanalytical method, a validation may be completed and reported with the long-term stability data appended to the report and analytical method summary, when appropriate.

Freeze–thaw stability

To assess freeze–thaw stability, at least three QC sample replicates are to be prepared at a minimum of two levels (ideally, the low- and high-QC concentrations) and subjected to at least three freeze–thaw cycles. Freeze–thaw QC samples are first frozen for a minimum of 24 h at the appropriate nominal storage temperature. Samples must be allowed to completely thaw unassisted at room temperature and then be refrozen for a period of at least 12 h per cycle. Acceptable thaw and refreeze time ranges are established prior to the onset of the freeze–thaw stability experiment, and documented accordingly during sample processing. It is recommended that the dilution QCs are also assessed for freeze–thaw stability.

Short-term, benchtop stability

QC samples are prepared in at least triplicate fashion for a minimum of two nominal concentrations (e.g., low- and high-QC levels) for assessment of short-term and benchtop stability. These QC samples are stored at room temperature for a time period sufficient to cover all sample handling activities (typically, 4–24 h). Additional QC samples may also be prepared for one or more unassigned time points in the event that unanticipated analytical difficulties are encountered.

Post-preparative stability

Processed sample stability is best described in the context of the incurred sample; namely, the validation experiment should mimic the stability of a processed incurred sample that is stored post-preparation on the autosampler and within the refrigerator or freezer prior to or following sample analysis. Since the post-preparative stability period is impacted by run length or reanalysis due to instrument or facility electrical failure, analysis of incurred samples (and respective QCs) is quantified against the same calibration standard curve that was prepared with the incurred samples and QCs. Therefore, it is a requirement that QC samples are reanalyzed against the original standard curve samples that were processed with the QCs. This stability experiment is frequently referred to as reinjection reproducibility. The process of reinjecting an accuracy and precision run and back-calculating the QC concentrations against a freshly prepared calibration curve (referred to as reinjection stability) is not a requirement, but it is recommended, and considered good practice (the FDA has issued form 483 for not demonstrating stability by analyzing QCs against a fresh curve).

To assess stability in the event of an unanticipated instrument or facility power failure; for example, a power failure over the weekend (72 h), QC samples are typically stored at autosampler conditions (in a tray at room temperature) for 48 to 96 h. In addition, one or more sets of prepared QC samples can be stored for a period of time sufficient to cover extract storage before analysis under appropriate conditions (e.g., 1-week refrigerator temperature extract storage). Although the latter practice is not required, it is also suggested that one of the initial accuracy and precision runs be retained on the autosampler, or at autosampler conditions, and reinjected after the desired stability period. In doing so, both reinjection reproducibility and reinjection stability can be assessed where the reinjected QC concentrations can be back-calculated using the original standard curve regression for stability assessment.

Selectivity

Selectivity is defined as the ability to differentiate and quantitatively determine the concentration of an analyte in the presence of other matrix constituents. Ideally, assessment of selectivity will be made prior to the onset of formal method validation experiments as part of method development. To formally demonstrate selectivity, blank control matrix from six or more independent sources should be obtained from different human or animal subjects and screened for potential interferences. The response in the blank matrix should be no more than 20% of the analyte signal at the LLOQ. It is acceptable if 80% of the selectivity samples meet acceptance criteria. Since the selectivity assessments performed in matrix are similar to incurred sample reanalysis (ISR), it may be desirable to procure additional sources of control matrix to capture potential variability in the endogenous analyte across several subjects. Special attention should be given to results that meet acceptance study criteria, but would potentially contribute to QC failures during accuracy and precision testing.

Proper homogenization of the control matrix from each source is a critical component of the sample preparation process. Trace elements are often endogenous to the biological matrix under investigation, in contrast to small organic molecules or xenobiotic compounds. Endogenous trace element levels can vary significantly between samples, or even within an organ from one subject, for a particular biological matrix. In addition, endogenous trace element concentrations can vary significantly as a function of diet and other exposure factors, so selection of appropriate control matrix and thorough homogenization is of critical importance. When quantifiable, endogenous analyte is detected, tissue homogeneity should be thoroughly evaluated, ideally as a component of method development. Following a tissue homogenization procedure, assessment of homogeneity can also be assessed by sampling and analyzing several aliquots of the homogenate.

The impact of any endogenous analyte on quantification should also be considered, especially when preparing matrix standards and QC samples. If recovery of the added analyte behaves in an additive, linear manner, then a correction factor may be applied, or background subtraction or orthogonal methodology, may be considered. The latter method requires the analysis of blank control samples so the endogenous level may be subtracted.

If analyte resolution from matrix interferences is not possible, their impact on selectivity can be determined through preparation and analysis of a matrix sample containing the interfering chemical species and the IS at their highest anticipated concentrations. This might be desirable if the likely presence of another endogenous element could potentially result in spectral or mass interferences during bioanalytical sample analysis. An interference check sample can be analyzed to determine the extent of the contribution of the interfering species to the analyte signal. The response for the interfering species will preferentially be no more than 20% of the response at the analyte LLOQ. If it is, then the impact of raising the LLOQ and/or changing the acceptance criteria on the method and incurred sample analysis must be considered.

Matrix effect

FDA guidance [101] indicates that, in the case of LC–MS- and LC–MS/MS-based procedures, matrix effects should be investigated to ensure that precision, selectivity and sensitivity will not be compromised. The matrix effect can be quantitatively expressed as the matrix factor (MF), defined as the ratio of the analyte peak response in the presence of matrix to the analyte response in the absence of matrix. An MF of 1 indicates no matrix effect, while an MF <0.85 or >1.15 indicates matrix suppression or enhancement, respectively [3]. For ICP-OES and ICP-MS measurements, matrix effects often stem from solution viscosity differences, sample solutions containing more total dissolved solids than the solvent, or from changes in analyte ionization efficiency as a function of the matrix. These differences can be mitigated through use of more extensive sample preparation procedures for matrix decomposition and/or the use of an appropriate IS that mimics the behavior of the analyte in the argon inductively coupled plasma.

One of the method validation considerations not documented in the 2001 guidance that has drawn industry opinion is with respect to the impact of hemolysis and lipidemic effects on a method, and whether these must be tested during method validation. The authors share the current consensus that these effects are validated during the matrix effect study, and investigations into the potential impact of hemolysis or lipidemic effects should be conducted during method development by investigating at least six lots of matrix including hemolyzed, hyperlipedemic and, if applicable, matrix from special populations such as renally or hepatically impaired populations [102]. Although this implies that separate method validation studies for these effects are not necessary, they are recommended if it is suspected that hemolytic and/or lipidemic effects from incurred samples may impact the data. This must be caveated with the potential that the European Medicines Agency, like Brazil, may require the evaluation of these effects during method validation, and therefore, if the intent is to validate a method that is globally acceptable, an understanding of the relevant current guidance is prudent.

Recovery

Recovery is defined as the ratio of detector response (or IS - corrected response) for an analyte in digested/extracted matrix to the response (or IS - corrected response) from the same amount of analyte in a non-digested/non-extracted, neat matrix [2], and can be expressed as:

For trace element determinations, unextracted samples are typically the dilute acid or acid mixture that was used during sample extraction, fortified with the analyte. Six or more replicates of each condition are prepared at each matrix QC concentration. For total trace element determinations, extraction or digestion is expected to be quantitative, with recoveries typically exceeding 90% with a high degree of precision (CV ≤ 15%). Lower recoveries may be an indication of an inefficient extraction or digestion. The determination of recovery differs from assessment of matrix effect in that for the latter, extracted/digested blank control matrix is fortified with the analyte after digestion.

Dilution integrity

Multiple replicates (at least five) of matrix QC samples should be prepared at the highest anticipated bioanalytical sample concentration for assessment of dilution integrity for sample concentrations exceeding the ULOQ. These samples should be diluted to approximately the middle of the calibration range with appropriate blank matrix. For trace element determinations, blank biological matrix extract or extract solvent are typically used for sample dilution, depending on the analyte of interest, potential matrix effects and endogenous analyte levels in the biological matrix of interest. In order for dilution integrity to be considered acceptable, the %RE and %CV must be within ±15% of the nominal concentration and less than 15%, respectively.

Contamination control

Although not strictly a component of method validation, it is recommended that the analytical and biological sample collection facilities collaborate to assess possible trace element contamination sources before sample collection [4,6]. Depending on the analyte of interest and method sensitivity, the use of plastic, glass, colored caps or containers, or selection of an improper anticoagulant could result in contamination. Method reagents should also be screened for analyte background prior to the onset of validation and bioanalytical activities. Once a suitable batch or lot number of a reagent has been identified, it may be desirable to procure enough material to ensure that alternate lots with potentially different analyte background levels are not used during the course of an investigation.

System suitability

System suitability is defined as the checking of a system to ensure that it is operating properly and in a manner consistent to achieve the required level of analytical performance. Successful system suitability is critical for the accurate, reproducible determination of trace elements in bioanalytical samples. Although suitability tests can be incorporated within an analytical run or prior to a run as part of an unattended autosampler sequence, it is recommended that suitability be successfully demonstrated and approved by a qualified analyst before the loading and analysis of all validation or bioanalytical samples. It is not required, nor is it recommended, that the acceptance of an analytical run be dependent upon the results of a system suitability test.

For ICP-OES and ICP-MS instruments, appropriate system suitability tests are often recommended by instrument manufacturers. These tests can be incorporated into instrument or project-specific standard operating procedures. A common approach to system suitability is analysis of a performance solution containing an element or a suite of elements at specified concentrations. The resulting atomic emission (ICP-OES) or mass spectral (ICP-MS) profiles must achieve predetermined sensitivity, stability and selectivity criteria. For example, ICP-MS system suitability elements (⁹Be, ⁵⁹Co, ⁸⁹Y, ¹¹⁵In and ²⁰⁹Bi) and background points (⁵AMU, ¹⁰⁰AMU, ¹⁵⁰AMU and ²²⁰AMU) can be selected to cover the majority of the mass range or to be free of interferences.

Failed ICP-OES or ICP-MS system suitability is often attributable to sample introduction system problems (probe, pump tubing, nebulizer or spray chamber). In contrast to small-molecule chromatographic measurements, the most common mode of sample introduction for ICP-OES and ICP-MS is direct nebulization of a liquid sample following an extraction or digestion procedure. If a thorough investigation of the introduction system fails to identify a problem, other parameters (e.g., forward power, lens settings, detector voltage) should be examined. As previously described, many parameters that influence system suitability are considered tunable and should be optimized to achieve the best possible performance from the instrument, while other parameters should be set in the method validation and maintained throughout sample analysis. Only minor adjustments to these tunable parameters are typically needed to achieve successfully system suitability. However, large changes to any tunable setting or multiple settings may indicate an instrument problem. For a summary of general tunable and non-tunable ICP-OES and ICP-MS instrumental parameters, please refer to Table 2.

Partial versus cross-validation

A partial validation is a modification of a previously validated bioanalytical method, which can range from as little as one intra-assay accuracy and precision determination to a nearly full validation. The partial validation is an extension of the respective full validation where an additional experiment(s) is conducted to ensure that the impact of a change in a method parameter(s) does not impact the reliability and robustness of the method, or data integrity when alternate method parameters are used with the same method. Reliability and data impact are determined using the same experimental criteria originally validated for any applicable experiment, and always require at least one accuracy and precision experiment to demonstrate that there is no impact on the calibration curve and QCs.

In order to determine which validation experiments are to be conducted to qualify the method revision, the impact of the change of the method parameter(s) on the properties of method must be determined, and the relevant validation experiment that measures that property conducted. For instance, a change in pump tubing size would require an accuracy and precision run as well as an assessment of carryover, but a change in the digestion mixture would best require a single accuracy and precision run, with LLOQ-QC qualification, reinjection reproducibility, recovery, matrix effect and carryover analysis to address all aspects of the method that could be affected. If partial validation experiments meet a priori defined and validated acceptance criteria, then the method parameters can be used with confidence in the revised method. Table 3 describes many examples of method parameter changes and suggested method validation experiments that should be considered to adequately assess the impact on the validated method.

It is important to note that changing instrument non-tunable parameters constitutes a necessity for a partial validation since the non-tunable parameters were validated. Moreover, if a partial validation fails and method parameter changes are necessary, then a full validation would be recommended. A change in the range of a calibration curve or the addition of new QC levels would require an accuracy and precision run before these could be used in the respective method. The partial validation data are usually appended to the validation report, and the analytical method summary revised (and versioned if associated with a GLP study), accordingly.

A cross-validation is a comparison of validation parameters when two or more bioanalytical methods are used to generate data within the same study or across different studies. This may include the bioanalytical instance where the original validated method serves as the reference and the transferred method is the comparator. Note that the comparisons should be conducted both ways. A cross-validation should also be considered when data are generated using different analytical techniques (e.g., ICP-MS, ICP-OES or atomic absorption) in different studies that are included in a regulatory submission, or when data from the same study are obtained across different instruments, including a different brand and variation of the same type of analytical instrument. An example of a cross-validation being necessary as a result of instrument or brand change would be the transfer of a method from a PerkinElmer ELAN 6100 to a PerkinElmer ELAN DRCII with Axial Field, or to an ICP-MS analytical instrument within or between two or more different laboratory facilities. A cross-validation is typically the same as a full validation except that the determination of matrix stability, including freeze–thaw stability, is theoretically not necessary since it has been validated previously. Many bioanalytical contract research facilities are nonetheless conducting matrix stability despite being provided necessary documentation from the original laboratory as an assurance of these critical method parameters.

The cross-validation should be performed and meet applicable acceptance criteria prior to the analysis of incurred samples on the instrument and/or within the facility for which the method cross-validation occurred. When sample analyses within a single study are conducted at more than one site or laboratory (method transfer), the FDA expects that a cross-validation with spiked calibration standards and subject samples (or qualified QCs should incurred samples not be available) should be done at each site or laboratory to establish inter-laboratory reliability. The authors recommend that the same set of QC samples be evaluated at both laboratories (original and the one where the method is being transferred to), and the difference between the QC measurements are to be the same as the QC acceptance criteria described in the original method, with CV duplicates <15%, and slopes between 0.85 and 1.15.

Method transportability

With increasing globalization of drug-development activities, possible transfer of the validated bioanalytical method to an alternate laboratory should be considered during the validation process [9]. To facilitate transfer of a trace element analysis method, emphasis should be placed on development of robust sample preparation and analysis procedures during method development. This is especially critical if the method is ultimately to be transferred to a laboratory with ICP-OES or ICP-MS instrumentation from a different manufacturer or platform as the system used during validation. For example, ion optic and collision cell configurations (hexapole collision cell, dynamic reaction cell, etc.) for quadrupole-based ICP-MS instruments often differ significantly between instruments, even from the same manufacturer. A validation may be successfully transferred to another laboratory with alternative instrumentation if sample preparation and analysis parameters are judiciously selected. In some instances, the eventual transfer of a method may preclude the use of instrumentation that is not readily accessible in the sample analysis laboratory (e.g., high-resolution, magnetic sector field ICP-MS).

Conclusion

Trace elements are critical components of many biological systems and therapeutic agents, but there are currently limited data related to the validation of methods for their determination in biological matrices. As a result, method validation practices are not always standardized across the bioanalytical support industry as laboratories attempt to account for issues specific to trace element measurements in different ways. Although the approach for method validation is essentially the same for trace elements and for small-molecule organic compounds, there are some important differences for bioanalytical method validation for trace elements that stem from their physiochemical properties and their most common detection modalities, such as ICP-OES and ICP-MS. For example, because of the relatively ubiquitous nature of many trace elements, special attention to controlling the analyte contribution from the reagents and the analytical procedure is often recommended. Trace elements are often endogenous to biological matrices, and normal levels within a matrix can vary significantly between samples. As a result, attention to identifying and achieving adequate homogenization of a suitable control matrix is critical.

The total concentration of a trace element is generally stable in biological matrix under varying storage conditions, while the stability for different species of the same element can vary significantly. As a result, potentially different volatility, precipitation and container-wall adsorption characteristics should be noted during method development. If the recovery is not adequate for a total trace element determination, a more rigorous sample preparation approach can often be utilized. Although ISR is an aspect of method validation [19], the authors feel that the discussion of when ISR should be conducted with elemental analysis is beyond the scope of this paper.

For ICP-OES and ICP-MS analyses, significant bias is generally not observed over long analytical runs, allowing for the use of one calibration curve at the onset of an analysis, provided that acceptance criteria are clearly defined a priori. Simple, unweighted linear regressions generally provide acceptable accuracy and precision over long concentration ranges, and NIST-traceable standards can be used for the preparation of matrix standards and QC samples. These solutions have established stability profiles, and their use, where practical, can facilitate the validation process.

Future perspective

It is becoming increasingly apparent that trace elements are central to many biological processes, and that subtle concentration changes within an organism from exposure or genetically predetermined metabolic tendencies can have a profound impact on the genesis, prevention or treatment of disease. As the knowledge base about the biological importance of trace elements continues to grow, the development of novel therapeutic agents or identification of new biomarkers based on trace elements will be likely to increase. Novel technologies or markers will require accurate, reproducible determination of trace elements in biological matrices to ensure the ultimate success of pharmacokinetic, bioequivalence, toxicological or epidemiological studies. Although many atomic spectrometric techniques are available for determination of trace elements, ICP-OES and ICP-MS will likely be the most relevant modes of instrumental analysis for bioanalytical method validation and sample analysis. These powerful techniques provide a high level of sensitivity while offering rapid, multielement analyses that are relatively free from significant interferences. In addition, both ICP-OES and ICP-MS can be interfaced with chromatographic instrumentation to provide trace element speciation data. The separation and measurement of the chemical form, oxidation state, or macromolecular binding of a trace element is critical to understanding its likely biological impact, so speciation studies are likely to play an increasingly larger role in bioanalytical sample analysis as the technology for conducting these measurements continues to improve. Regardless of the atomic spectrometric technique, an increasingly global economy will place greater emphasis on the development and validation of rugged bioanalytical methods that are compliant with harmonized guidance and are readily transferable to CROs in several countries.

Table 1. Summary of fundamental and supporting trace element validation parameters.

Validation test	Runs	Criteria
Precision and accuracy	Three	±/≤ 15/20% (refer to QCs, below)
Standard curve	All	±/≤ 20% at LLOQ-STD; ±/≤ 15% all other STDs; 75% pass
QCs		Intra- and inter-assay accuracy (%RE) ±15% and precision is ≤15%, except at the LLOQ QC (±20 and ≤20%, respectively); n = 6 in the three P&A runs
LLOQ	Three	±/≤20%
Inter-lot matrix selectivity	One run ≥ six lots	≤20% LLOQ in ≥80% of lots; run meets passing criteria
Inter-lot accuracy: matrix spike at LLOQ and ULOQ	One run ≥ six lots	±15%; LLOQ ± 20%; ≥80% of lots do not exhibit internal standard response in matrix blanks; run meets passing criteria
Matrix effect	One run three levels	Matrix factor consistent; n ≥ 6 for each level; CV ≤ 15%
Extraction (digestion) recovery	One run three levels	% recovery consistent; CV ≤ 15%; n ≥ 6 for each level; rejection of one replicate sample only
Carryover	All	≤20% mean LLOQ standard response area of analyte/IS response
Benchtop stability	One	Include: 0 h, ≥4 h and 24 h; low-QC, mid-QC, high-QC, Dil-QC (n ≥ 3); run meets passing criteria
Freeze–thaw stability	One	≥3 cycles; low-QC and high-QC (n≥3); run meets passing criteria
Long-term stability -20°C (range = SOP-dependent)	1 month 3 months	Precision and accuracy: ±/≤ 15%/20%; refer to QCs
Reinjection reproducibility	One	Precision and accuracy: ±/≤ 15%/20%; refer to QCs and standard curve
Run length robustness	One	Validate and document run sample size
Analyst robustness	One	Precision and accuracy is determined over at least 2 days with ≥1 different analyst
System suitability	All	Before each run; not as part of run

CV: Coefficient of variation; QC: Quality control; SOP: Standard operating procedure.

Table 2. Typical tunable and non-tunable ICP-OES and ICP-MS instrumental parameters.

ICP-OES
Tunable	Nontunable
Torch position	Pump tubing size
Nebulizer flow and pressure	Peristaltic pump rate (sample introduction rate)
Detector voltage	Spray chamber type and temperature
	Rinse solution: composition, time, pump rate
	Analyte and internal standard emission wavelengths
	Inductively coupled plasma radiofrequency power
	Auxiliary and coolant plasma gas flows
	Read delay and integration times
	Axial or radial viewing configuration
	Reading/replicates
ICP-MS
Tunable	Nontunable
Torch position	Pump tubing size
Nebulizer flow and pressure	Peristaltic pump rate (sample introduction rate)
Ion optic lens voltages	Spray chamber type and temperature
Collision or reaction cell gas flow	Rinse solution: composition, time, pump rate
Detector voltage	Analyte and internal standard masses
	Collision cell gas composition
	Inductively coupled plasma radio frequency power
	Auxiliary and coolant plasma gas flows
	Read delay and integration times
	Dwell time, channels/mass, sweep number
	Reading/replicates
	Interface cones: composition and geometry
	Interface cones: orifice size

ICP-MS: Inductively coupled plasma-MS; ICP-OES: Inductively coupled plasma-optical emission spectrometry.

Table 3. Partial validation conditions.

Validation parameter	Experimental requirements (minimum criteria)
Matrix (species-related)	Full validation
Plasma anticoagulant	Single A&P, inter-lot (≥ six lots) or transparency study
Autosampler	Single A&P, carryover
Extraction/digestion	Single A&P with LLOQ-QC, reinjection reproducibility, carryover
Rinse solution change: composition, rate	Single A&P, run length robustness, carryover
Sample introduction system: spray chamber, nebulizer	Single A&P, matrix effect, run length robustness, carryover
Sample introduction rate: pump tubing size, speed	Single A&P, matrix effect, run length robustness, carryover
Analyte or IS emission wavelength or isotope	Full validation (except matrix stability)
Viewing condition: axial versus radial	Full validation (except matrix stability)
ICP parameters: RF power; coolant; auxiliary flow ranges	Three A&P, single LLOQ-QC (n = 6)
ICP-MS collision cell gas composition	Three A&P, three LLOQ-QC (n = 6), inter-lot selectivity and accuracy (≥ six lots), matrix effects
ICP-MS cone interference	Three A&P, single LLOQ-QC (n = 6), inter-lot selectivity and accuracy (≥ six lots), matrix effects, run length robustness
ICP-OES or MS instrument platform	Three A&P, single LLOQ-QC (n = 6), inter-lot selectivity and accuracy (≥ six lots), matrix effects
IS	Single A&P, inter-lot selectivity and accuracy (≥ sixfold)
QC level change	Single A&P
Change in STD curve range (ie., LLOQ)	Single A&P, matrix stability study if concentration specifications not met with change
Post-validation stability	Single A&P or high- and low-QCs with run qualifying QCs
Curve fitting change	Single A&P, LLOQ-QC (n = 6), qualification new QC levels
Pump tubing size	Single A&P, LLOQ-QC (n = 6), carryover

ICP: Inductively coupled plasma; IS: Internal standard; OES: Optical emission spectrometry; QC: Quality control; RF: Radiofrequency.

Total trace element determinations

Quantitative measurement of trace elements in a sample, without regard to chemical form or oxidation state.

ICP-OES

Inductively coupled plasma-optical emission spectrometry. Rugged, mature analytical technique for the determination of trace elements in a variety of matrices at parts-per-million (µg/ml) and parts-per-billion (ng/ml) levels.

ICP-MS

Inductively coupled plasma-MS. State-of-the-art, reliable analytical technique for the determination of trace elements in a variety of matrices at parts-per-billion (ng/ml) and parts-per-trillion (pg/ml) levels.

NIST-traceable standards

Commercially available trace element reference standards that are traceable to National Institute of Standards and Technology (NIST) standards.

Trace element speciation

The separation and quantitative measurement of the chemical form or oxidation state of a trace element.

Ethical conduct of research

The authors state that they have obtained appropriate institutional review board approval or have followed the principles outlined in the Declaration of Helsinki for all human or animal experimental investigations. In addition, for investigations involving human subjects, informed consent has been obtained from the participants involved.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.