Assessment of matrix effect in quantitative LC-MS bioanalysis

LC-MS, predominantly with electrospray ionization (ESI) source, has been extensively employed for the determination of drugs and/or their metabolites in various biological matrices (e.g., plasma, serum, blood, urine and tissues) in support of preclinical and clinical development. However, the determination is often subject to matrix effect due to the presence of matrix components coeluting with the analyte of interest. The matrix effect-introduced MS signal suppression or enhancement may lead to erroneous results. Therefore, a solid matrix effect assessment is essential during method development to understand the possible impact on method performance and elucidate appropriate mitigation prior to method validation. The current editorial summarizes the best practices of matrix effect assessment and associated corrective measures to ensure the robustness of the intended LC-MS bioanalytical method.

Matrix effect in LC-MS bioanalysis

Matrix effect is one of the key parameters of a given LC-MS bioanalytical method. It refers to the adverse impact caused by components coeluting with the analyte of interest [1–5]. The components could be endogenously present in the matrix, for example, phospholipids, proteins and salts; or exogenously introduced to study samples, for example, anticoagulants, dosing vehicle, stabilizer and co-medication. The impact of matrix effect on LC-MS bioanalysis varies depending on its origin, extent and the intended method. Sometimes matrix effect is caused by non-resolved chromatographic peak with the same MS/MS transition as the target analyte. This could be resolved by modifying LC method or incorporating other techniques (e.g., ion mobility spectrometry) [4]. More often, matrix effect is reflected by a significant change in ionization efficiency (signal enhancement or suppression) of the target analyte [6]. Unfortunately, this defect is often undetected by simple examination of LC-MS chromatograms and, if not properly interrogated and addressed, could lead to suboptimal method performance (e.g., poor accuracy and precision, nonlinearity and reduced sensitivity, etc.), particularly when internal standard (IS) does not properly track the analyte during LC-MS bioanalysis.

Matrix effect assessments

Matrix effect can be assessed qualitatively and/or quantitatively via post-column infusion, post-extraction spiking and pre-extraction spiking.

Post-column infusion

A constant flow of analyte neat solution is continuously introduced (commonly via a syringe pump) into and mixed with the post-column eluent of an injected blank matrix extract before the eluent enters the MS system. Ion chromatogram for the analyte is monitored. Any significant disruption (increase or decrease) of the MS signal for the analyte in the ion chromatogram indicates ion enhancement or suppression. Although the approach does not give quantitative details of matrix effect, it is of great value during method development and troubleshooting as it allows for obtaining information on the region and extent of ion enhancement/suppression, if any, throughout the LC-MS run. Additional effort can be made by phospholipid monitoring to assess whether the observed matrix effect is due to endogenous phospholipids or not [7]. Accordingly, the intended chromatographic method and/or sample preparation procedure can be modified to ensure analyte is less prone to matrix effect.

Post-extraction spiking

The method was introduced by Matuszewski et al. [1] and has been adopted as a ‘golden standard’ to quantitatively assess matrix effect in regulated LC-MS bioanalysis. Briefly, the method involves calculation of LC-MS response ratio (also called matrix factor [MF]) of analyte and/or IS spiked in the post-extraction blank matrix versus in the neat solution(s) at the corresponding concentration(s). MF of <1 suggests a signal suppression while >1 a signal enhancement. By calculating MF, possible lot-to-lot variation and/or concentration dependency of matrix effect could be assessed. In authors' organization, a combination of post-column infusion and post-extraction spiking has been utilized to effectively guide method development and optimization to identify and remove, if possible, matrix effect.

It is worth noting that a complete removal of matrix effect sometimes is challenging and time consuming. Alternatively, the focus could be shifted to compensation for the matrix effect by employment of a suitable IS [5,8]. A proper IS with good trackability is expected to experience a similar or the same matrix effect as the target analyte, which is demonstrated quantitatively by the IS-normalized MF (calculated as MF of the analyte/MF of the IS) being close to one. With this regard, stable isotope labeled (SIL) IS, such as ¹³C-, and ¹⁵N-labeled, is considered the best choice in LC-MS bioanalysis, as it co-elutes with the analyte.

Pre-extraction spiking

As captured in the ICH M10 guidance [9], the method focuses on the evaluation of accuracy and precision of low and high QCs prepared in different sources/lots of matrix blank (e.g., at least six different matrix lots, hemolyzed and/or lipemic matrix). The acceptable results (bias within ±15% and CV ≤15% in each individual source of matrix) of the pre-extraction spiked QCs (low and high) serve as a qualitative demonstration of a consistent matrix effect, if any, of the method. However, the results provide no information on the scale of matrix effect (signal enhancement or suppression) for the needed troubleshooting when issue occurs. In a case at authors' organization, significant signal enhancement was observed for both the analyte and SIL IS as evidenced by the absolute MFs being greater than 3. Although the pre-spiked QC results at low and high concentration levels met the acceptance criteria (suggesting matrix effect was compensated for by SIL IS), a lot-dependent signal enhancement may result in large variation in IS responses of incurred samples. Upon investigation and method modification, the issue was resolved by switching ionization mode from ESI to atmospheric-pressure chemical ionization (APCI), which is less susceptible to matrix effect.

Practical recommendations

In method development, every effort should be made to optimize sample extraction method and LC conditions to ensure a consistent MF for the target analyte across the low and high QC concentration levels. In authors' view, for a robust LC-MS bioanalytical method, the absolute MFs for the target analyte via post-extraction spiking approach should be, ideally between 0.75 and 1.25 and non-concentration dependent. IS normalized MF should be close to 1.0, regardless of whether SIL or analogue IS is employed. Once matrix effect is noticed, effort should be made to remove it and/or mitigate its impact. Common approaches include but are not limited to additional sample cleanup, chromatographic separation of the interfering components and employment of a different ionization mode, such as APCI [10–14]. While switching from ESI to APCI might be straightforward in mitigating matrix effect, APCI has its own limitations, for example, not suitable for highly volatile analytes, sensitivity limitation [15].

During method validation, matrix effect should be evaluated confirmatively by analyzing low and high QC samples (at least 3 replicates at each level) prepared in at least six different sources/lots of matrix blank, whenever possible, as well as in hemolysed and/or lipemic matrix blank [9]. Accuracy and precision of QC results should be within the set criteria (i.e., within ±15% bias and ≤15% CV) to confirm that the matrix effect, if any, has no impact on method performance. Evaluation of absolute and IS normalized MFs might be skipped if already assessed during method development with satisfactory outcomes.

It is worth noting that the matrix components in incurred samples are much more complex than the blank matrix used for the preparation of calibration standards and QCs owing to the presence of dosing vehicle, subject-specific endogenous components, drug metabolite(s), co-administered drugs and their metabolites, etc. [16]. This may lead to unattended errors in the measured concentrations, although the results of calibration standards and QCs might meet the acceptance criteria. Consequently, monitoring IS responses becomes critical during sample analysis. For samples with abnormal IS responses, repeat analysis with a dilution factor greater than the initial analysis is recommended. If the IS responses of the diluted samples come back to normal, it indicates that the unexpected matrix effect exists in the study samples. If analyte concentrations obtained from this repeat analysis are within ±20% (difference) of the original values, the study sample specific matrix effect can be judged to have no impact on the measurements. For studies anticipating matrix effect (i.e., intravenous administration with dosing vehicle containing PEG-400 or Tween-80), it is recommended to pre-dilute study samples collected at early time points (i.e., up to 4 h), if sensitivity is not an issue.

Summary

In summary, matrix effect can be assessed qualitatively by post-column infusion and pre-extraction spiking, and quantitatively by post-extraction spiking during LC-MS bioanalytical method development and/or validation to systematically understand the presence of matrix effect, if any, and its possible impact. Failure in timely investigating and mitigating the impact due to matrix effect could be detrimental. Matrix effect can be effectively compensated for by employment of a suitable IS. Even with the latter, necessary effort is still necessary during method development to reduce or totally remove matrix effect by optimizing sample extraction method and LC chromatographic conditions to ensure the method robustness. It is highly recommended to monitor IS responses during sample analysis to spot subject-specific matrix effect, if any, for necessary investigation and mitigation. Pre-dilution of study samples with anticipated matrix effect is also a good practice to mitigate the possible impact in advance.

Financial disclosure

The authors have no 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.

Competing interests disclosure

The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, stock ownership or options and expert testimony.

Writing disclosure

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