Tutorial review on validation of liquid

Analytica Chimica Acta 870 (2015) 29–44
Contents lists available at ScienceDirect
Analytica Chimica Acta
journal homepage: www.elsevier.com/locate/aca
Tutorial
Tutorial review on validation of liquid chromatography–mass
spectrometry methods: Part I
Anneli Kruve a , Riin Rebane a , Karin Kipper a , Maarja-Liisa Oldekop a , Hanno Evard a ,
Koit Herodes a , Pekka Ravio b , Ivo Leito a, *
a
b
University of Tartu, Institute of Chemistry, Ravila 14a, 50411 Tartu, Estonia
Finnish Customs Laboratory, Tekniikantie 13, PL 53, FI-02151 Espoo, Finland
H I G H L I G H T S
G R A P H I C A L A B S T R A C T
The status of validation of LC–MS
methods
is
comprehensively
reviewed.
Clarity is brought into validationrelated terminology.
Recommendations on difficult validation-related issues in LC–MS are
given.
A R T I C L E I N F O
A B S T R A C T
Article history:
Received 4 June 2014
Received in revised form 31 January 2015
Accepted 9 February 2015
Available online 13 February 2015
This is the part I of a tutorial review intending to give an overview of the state of the art of method
validation in liquid chromatography mass spectrometry (LC–MS) and discuss specific issues that arise
with MS (and MS/MS) detection in LC (as opposed to the “conventional” detectors). The Part I briefly
introduces the principles of operation of LC–MS (emphasizing the aspects important from the validation
point of view, in particular the ionization process and ionization suppression/enhancement); reviews the
main validation guideline documents and discusses in detail the following performance parameters:
selectivity/specificity/identity, ruggedness/robustness, limit of detection, limit of quantification, decision
limit and detection capability. With every method performance characteristic its essence and
Keywords:
Liquid chromatography–mass spectrometry
Abbreviations: APCI, atmospheric pressure chemical ionization; API, atmospheric pressure ionization; APPI, atmospheric pressure photo ionization; ban, the slope of the
calibration function for the analyte; bint, the slope for the potential interferent; CCa, decision limit; CCb, detection capability; EMA/EMEA, European Medicines Agency; ESI,
electrospray ionization; Ex, effect of variation of parameter X; FDA, United States Food and Drug Administration; HRMS, high resolution mass spectrometer; ICH, International
Conference on Harmonization; IR, infrared spectroscopy; IUPAC, International Union of Pure and Applied Chemists; LC–MS, liquid chromatography–mass spectrometry; LLE,
liquid–liquid extraction; LoD, limit of detection; LoQ, limit of quantitation; LRMS, low resolution mass spectrometer; MEionization, matrix effect of ionization; MRM, multiple
reaction monitoring; MS/MS, tandem mass spectrometry; MSn, consequent reaction monitoring; NLS, neutral loss scan; NMR, nuclear magnetic resonance; Rs,
chromatographic resolution; RSD, relative standard deviation; s, standard deviation; SPE, solid phase extraction; SRM, selected reaction monitoring; TOF, time-of-flight;
UPLC/UHPLC, ultra-high performance liquid chromatography; UV–vis, ultraviolet–visible spectrophotometry; VIM, International Vocabulary of Metrology; a, probability of a
blank sample being considered as a positive sample; b, probability of falsely negative result.
* Corresponding author. Tel.: +372 7 375 259; fax: +372 7 375 264.
http://dx.doi.org/10.1016/j.aca.2015.02.017
0003-2670/ ã 2015 Elsevier B.V. All rights reserved.
30
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
Validation
Electrospray
Ionization suppression
Limit of detection
Robustness
terminology are addressed, the current status of treating it is reviewed and recommendations are given,
how to determine it, specifically in the case of LC–MS methods.
ã 2015 Elsevier B.V. All rights reserved.
Contents
1.
2.
3.
4.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
MS as a detector for LC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.
Ionization sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Ionization suppression or enhancement effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.
Operation modes of MS as a detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.
Practical aspects of LC–MS method development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4.
Overview of the main validation guidelines and general situation with validation of LC–MS methods
Parameters of LC–MS methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1.
Selectivity, specificity, confirmation of identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Selectivity (specificity) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1.1.
Identity confirmation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1.2.
Ruggedness/robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.
4.2.1.
Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Specific aspects in LC–MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.2.
Experimental design of ruggedness/robustness testing . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.3.
4.2.4.
Method parameters to be varied . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Numerically expressing robustness/ruggedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.5.
Limit of detection, limit of quantitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.
4.3.1.
Limit of detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Limit of quantitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.2.
Decision limit (CCa) and detection capability (CCb) . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.3.
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
31
32
32
32
33
34
34
34
34
35
35
36
36
37
37
37
38
39
39
40
41
41
41
Anneli Kruve obtained her Ph.D. from University of
Tartu (UT) in 2011. Since 2005 she has been involved
in HPLC and LC/MS method development and
validation in various fields: bioanalysis, food analysis as well as industrial analysis. Starting from
2011 she works as a research fellow at UT Institute of
Chemistry. In 2008 and 2009 she has worked in
University of Helsinki in the field on miniaturization
of MS ion sources. Her main research fields are
method development for pesticide analysis, fundamental studies of ionization efficiency and design of
MS ionization sources.
Riin Rebane obtained her Ph.D. in analytical
chemistry from University of Tartu in 2012 with a
topic on optimization and validation of liquid
chromatographic methods with mass spectrometric
detection containing derivatization. For the past
eight years her main research area has been LC–MS
analysis, including method development and validation for various analytes and development of
novel derivatization reagents for LC–MS. She also
works as a quality assurance specialist in the
Estonian Environmental Research Centre.
Karin Kipper obtained her Ph.D. from University of
Tartu (UT) in 2012. Since 2004 she has been involved
in the bioanalytical method development and
validation for HPLC-UV/vis and LC–MS analysis,
working at UT Institute of Pharmacology and
Institute of Chemistry. Starting from 2012 Karin
Kipper works as a research fellow at UT Institute of
Chemistry. Her main research fields are pharmaceutical bioanalysis (pharmacokinetic/pharmacodynamic studies), pharmaceuticals’ pathways in
environment and development of novel eluent
additives for LC–MS in order to improve separation
and peak shape of basic compounds.
Maarja-Liisa Oldekop obtained her M.Sc. from
University of Tarty (UT) in 2013 and is now a Ph.
D. student in the chair of analytical chemistry at UT.
Her main field of research is development of LC–MS
methods using derivatization. The focus is on matrix
influences on this type of analysis, stressing the
importance of trueness of the analysis results but
also the sensitivity of the method. Since the
beginning of 2013 Maarja-Liisa Oldekop works as
the quality manager of the UT Testing Centre, which
is an ISO/IEC 17025 accredited testing and calibration laboratory.
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
31
Hanno Evard graduated from the Applied Measurement Science master’s programme at University of
Tartu in 2012. His master’s thesis was about paper
spray ionization. He is now a Ph.D. student at
University of Tartu. His research is focused on
fundamental studies and developing new applications for different mass spectrometry ion sources.
Koit Herodes obtained his Ph.D. from University of
Tartu (UT) in 2002. Since 2008 he works as the head
of the UT Testing centre – a unit providing testing
and analysis services and accredited according to
ISO 17025 by the Estonian Accreditation Centre.
Since 2005 Koit Herodes works as associate professor of analytical chemistry at UT Institute of
Chemistry. He has been principal investigator of
numerous projects involving LC–MS analyses. Currently he is the principal investigator of the project
“Development of software for validation of chromatographic methods”, which aims at creating webbased software for validation of chromatographic
methods.
Pekka Ravio obtained his M.Sc. from University of
Helsinki in 1981. Since 1984 he has been working in
the Finnish Customs Laboratory being responsible
for various types of mass spectrometric analyses. His
main area of expertise is pesticide residue analysis
and quality assurance.
Ivo Leito obtained his Ph.D. from University of Tartu
(UT) in 1998. During 2002–2005 he reorganized the
UT Testing centre – a unit providing testing,
measurement, analysis and training services to
laboratories and industry – and was its director.
Starting from 2005 Ivo Leito works as professor of
analytical chemistry at Institute of Chemistry,
University of Tartu. His main research directions
are on the borderline of analytical chemistry with
other disciplines: chemistry of superacids and
superbases; metrology in chemistry (MiC); liquid
chromatography and mass spectrometry; sensors
and their metrological characterization; applications of instrumental methods in analysis of
historical objects. He teaches analytical chemistry
and its metrological aspects at UT and has been
involved in setting up several international MiC-related educational activities.
1. Introduction
During the recent 15 years liquid chromatography mass
spectrometry (LC–MS) has evolved from a scientific curiosity into
a routinely applied technique finding increasingly more use in
routine field laboratories. This has become possible largely due to
the advent of the atmospheric pressure ionization (API) methods
[1]. The API sources are able to produce gas-phase ions with little
or no spontaneous decomposition from delicate and high
molecular weight analytes. This, combined with the intrinsic
sensitivity of mass spectrometers, has revolutionized large areas of
chemical analysis where traces of organic analytes are determined
in complex matrices. Among the ionization methods electrospray
ionization (ESI) has proven especially versatile [1]. As a result,
almost all fields of chemical analysis (bioanalytical and medical,
environmental, food, drug discovery [2], etc.) have experienced big
changes.
LC–MS offers additional selectivity and confirmation of identity
compared to “traditional” chromatography (i.e., chromatography
with unidimensional detector, such as UV absorbance) by
determining the mass/charge ratio of the ion(s) or recording MS
data (in the broadest sense of this word) for the whole
chromatogram, often resulting in three-dimensional datasets.
This has proven immensely useful in the wide range of fields where
LC–MS is applied [1,3].
The extensive additional possibilities, however, come at a cost:
LC–MS systems are complex and a large number of parameters
have to be at or near optimal values in order to get the desired
performance [4,5]. This automatically means that whenever an
analytical method based on LC–MS is developed, its performance
has to be carefully checked and monitored. Nevertheless, the
instrument vendors are continuously making efforts to reduce the
complexity of the systems and their usage. In addition, the default
parameters are often sufficient for routine applications and the LC–
MS system software is increasingly user-friendly.
Method validation is a key activity in chemical analysis,
indispensable for obtaining reliable results [6]. The higher the
complexity of the method, the more important and voluminous, as
a rule, is validation [7,8]. Methods based on LC–MS are notorious
for their complexity, on the one hand because of the instrument
itself and on the other hand because LC–MS is often applied to the
most complex samples [5,9,10]. Besides the intrinsic necessity of
validation, there are increasingly more regulations affecting
laboratories that stipulate method validation as a requirement
[11–13] and scientific journals publishing analytical chemistry
research require validation data for the methods to be published. A
very large number of articles reporting new LC–MS methods are
submitted for publication each year and method validation is
usually an intrinsic part of such articles. However, in spite of the
number of articles submitted and published, miscalculation and
misinterpretation of validation parameters is still common due to
complexity of the methods and some ambiguity in the definitions
of some of the validation parameters [14].
The importance of validation has led to the emergence of a large
number of validation guidance materials for laboratories, both of
universal nature [15,16] and sector-specific [17–20]. Although
there is a general agreement on the various validation parameters
to be evaluated, diversity prevails about details and about the
methodology employed for validation and acceptance criteria
[21,22]: different recommendations and different sets of terminology are found in the different guidelines. As a conclusion, the
analytical community is still far from a consensus about exactly
how validation should be carried out, both in general terms and as
applied to LC–MS [14,23,24–25].
This tutorial review tries to overview these issues and to
address the validation of LC–MS methods. Most validation
guidelines do not specifically address LC–MS at all [15,26]. To
some others, LC–MS-specific issues have just recently been added,
especially to the ones targeted towards bioanalytical methods
[18,19]. Recently a review has been published on the validation of
32
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
bioanalytical chromatographic methods [27], which is relevant
also for LC–MS.
The specific aims of this tutorial review are threefold:
1. Give a critical overview of the state of the art of LC–MS method
validation especially with respect to the relevant guidance
materials;
2. Draw attention to the aspects that are important specifically in
validating LC–MS methods (e.g., ionization suppression by
matrix effects);
3. Give recommendations on problematic issues in validation of
LC–MS methods.
This review is written with highly pragmatic approach and is
expected to be useful besides scientific research laboratories also
for routine laboratories.
Few words on general terminology are appropriate here. For
measurement-related terms the definitions given in the International Vocabulary of Metrology (VIM) [28] are used in most cases.
The term method performance characteristic is used for e.g., limit
of detection (LoD), trueness, etc. The term method parameter is
used for adjustable parameters, such as mobile phase pH, flow rate,
column type, etc. The word technique is used in this tutorial for
describing the generic (instrumental) platform for analysis, e.g.,
LC–MS. The word method means an analytical methodology for
solving a specific analytical task (e.g., determination of a set of
drugs in a specific matrix, such as blood plasma, using LC–MS with
ESI ionization). In some documents [28] procedure is recommended in this meaning, but the term overwhelmingly used by the
analytical community is method and we choose to use it. The term
approach is used in the context of determining method performance characteristics, e.g., there are different approaches for
determining LoD, assessing linearity, etc. The term run refers to the
analysis of a single sample (i.e., a single chromatogram). The term
(analytical) sequence refers to a set of samples analysed in
sequence (e.g., with an autosampler) with the same method (i.e., a
series of chromatograms).
With every method performance characteristic (e.g., trueness,
limit of detection, . . . ) we explain its essence and terminology,
the current status of calculating and interpreting it and give our
recommendations, how to treat it, specifically in the case of LC–MS
methods. Validation is a complex and multifaceted activity, not
always easily separable from the process of analytical method
development. Therefore, occasionally our suggestions address also
method development. It is assumed that the system suitability has
already been verified, so system suitability checking is not
addressed here.
This tutorial review is divided in two parts. The Part I briefly
introduces the principles of operation of LC–MS (emphasizing the
aspects important from the validation point of view, in particular
the ionization process and ionization suppression/enhancement);
reviews the main validation guideline documents and discusses in
detail the following performance parameters: selectivity/specificity/identity, ruggedness/robustness, limit of detection, limit of
quantification, decision limit and detection capability. The Part II
[29] starts with briefly introducing the main quantitation methods
and then addresses the performance related to quantification:
linearity of signal, sensitivity, precision, trueness, accuracy,
stability and measurement uncertainty. Its last section is devoted
to practical considerations in validation.
Literature starting from year 2000 is mainly covered, but
occasionally earlier works of lasting significance have been
included. ESI as the most popular ionization method is covered
more thoroughly (and is the default ionization method), but most
of what is written also applies to the atmospheric pressure
chemical ionization (APCI) and atmospheric pressure photo
ionization (APPI) sources. Literature references are by no means
exhaustive. Preference has been given to literature sources that
either specifically focus on validation of LC–MS methods or where
some important aspect of LC–MS method validation is highlighted.
This review addresses LC–MS with analytes that are separated with
the conventional LC columns, i.e., small to medium-size organic
molecules (with molecular weight up to few thousands).
2. MS as a detector for LC
The success of the LC–MS technique arises from its ability to
give three-dimensional data. First, the compounds are separated in
time by LC. Ions generated in the ionization source are then
separated according to their m/z ratios in the mass analyzer of MS.
Finally, the MS detector measures the abundance of each ion.
Compared to the traditional LC detectors, such as ultraviolet–
visible spectrophotometry (UV–vis) or fluorescence, the MS
detector enables significantly more reliable identification of the
compounds eluting from LC. This is true for the conventional
nominal resolution MS (typical examples are triple quadrupole and
ion trap instruments), especially if MS/MS detection (see below) is
used, and even more so for the high-resolution instruments (such,
as time-of-flight (TOF) or Orbitrap). Therefore, among other
benefits, LC–MS reduces the risk of false positive identification.
Another advantage over other techniques is that difficult analytes
can be successfully determined in complicated matrices at low
levels.
2.1. Ionization sources
For a long time combining LC with MS was hindered by the
difficulties in transferring analyte molecules from the liquid phase
to the gas phase as ions. These difficulties were solved by the
invention of the API sources [30]. API sources enable formation of
gas-phase ions directly from liquid flow and thereby enable
connecting LC to MS. The most common API sources, and in the
focus in this paper, are electrospray (ESI) [31], APCI [32] and APPI
[33,34]. API sources cover a wide range of analyte and solvent
polarity.
Both positive and negative ionization mode are available in
most API sources, though positive mode is more extensively used.
Positive ions are usually formed as protonated analyte molecules
[M + H]+ but also adducts with cations (mostly sodium, ammonium
or potassium), e.g., [M + Na]+ [35] or, in less common cases, as
radical cations after loss of an electron M +.
For negative ionization mode, also different forms of ionization
may occur. Similarly to positive ionization, the most significant is
deprotonation [M H] . Less common is adduct formation with
e.g., chloride [M + Cl] , nitrate [M + NO3]
or acetate [M +
CH3COO] . Also radical anions via addition of electron may occur
[31].
The ionization in API sources is “soft”, meaning that the formed
ions have little excess energy and therefore undergo little or no
fragmentation. This is very different from the electron ionization
(EI), which is typically used in gas chromatography mass
spectrometry, where fragmentation (often extensive) is rather a
rule than an exception [1].
2.2. Ionization suppression or enhancement effect
In API sources, especially in ESI, ionization efficiency of the
analytes may be strongly altered by the compounds co-eluting
(matrix compounds either from the sample analysed or very late
eluting compounds from previous samples [36,37]) with the
analyte [38,40]. The effect may either reduce (called ionization
suppression) or increase (ionization enhancement) the analyte
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
33
signal. Often both suppression and enhancement effects are
abbreviated in the term matrix effect. The term matrix effect
can also be understood in a broader sense – “the combined effect of
all components of the sample other than the analyte on the
measurement of the quantity” [39]. It means any influence that
sample matrix may have on an analytical result and is applicable to
any analytical method. This paper focuses only on the narrow
meaning of the matrix effect – a factor affecting the ionization
process occurring in LC–MS ionization sources. Therefore, to avoid
confusion, we denote it as MEionization (as ME is often used in LC–
MS literature) and preferably use terms ionization suppression and
enhancement in text.
The ionization suppression may result in more than 90% of the
signal decrease but can also lead to the complete loss of MS signal
resulting in false negative results [38]. The mechanism of
ionization suppression is very complex and several reviews have
been published [41,42]. Therefore, the aim of this chapter is not to
summarize the available literature, but to describe the most
important aspects from the validation point of view.
Whenever possible and practical, ionization suppression should
be eliminated or significantly reduced. If it is not possible to reduce
ionization suppression to the level of being insignificant, it should
be taken into account in calculation of the results. Several
approaches [36,43–46] have been suggested and tested for
reducing the ionization suppression effect, mainly focusing on
ESI ionization source. In broad terms the approaches can be
categorized as (1) sample preparation based, (2) instrumental
modifications and (3) modifications in LC method:
some cases. Unfortunately, there are numerous analytes for
which neither APCI nor negative mode ESI are suitable.
(3) The two main LC-method-related matrix effect reduction
possibilities are improvement of chromatographic separation,
e.g., with ultra-high performance liquid chromatography
(UPLC/UHPLC), and sample dilution. Both have been used by
numerous authors [55–58]. Dilution has been shown to
significantly reduce the ionization suppression [56–58].
However, it is often impossible to dilute the sample sufficiently
so that ionization suppression will completely disappear,
because the analyte concentration may fall below the limit
of quantification. In such cases, the so-called extrapolative
dilution approach has been found useful [45], which consists in
diluting the sample as far as possible and if the suppression is
still present then extrapolating the analyte concentration
mathematically to infinite dilution.
(1) Less than ideal sample preparation may be viewed as the main
Intensive development of mass spectrometry instrumentation
has enabled an increasingly diverse range of possible operation
modes [1]. This has led to the situation that nowadays the majority
of analyses carried out with LC–MS use tandem mass spectrometric (often abbreviated as MS/MS) detection. In very broad terms,
MS/MS means that the ions obtained from the sample are not
directly detected to obtain the signal but, depending on purpose,
some of them (called precursor ions) are selected from the
spectrum, manipulated (usually fragmented) and the resulting
ions (product ions) are eventually detected. MS/MS, as opposed to
conventional MS, is popular in routine analysis because it enables
more reliable identification of the analyte (in the sense of lower
false positive rate) and higher signal-to-noise ratio.
MS/MS is defined as “acquisition and study of the spectra of the
product ions or precursor ions of m/z selected ions, or of precursor
ions of a selected neutral mass loss” [61] and embraces several
operation modes. Different MS operation modes such as selected
reaction monitoring (SRM), multiple reaction monitoring (MRM),
consequent reaction monitoring (MSn) and neutral loss scan (NLS)
are frequently used. Even though mass spectrometer operation
mode does not change the general validation rules, some
modifications may be needed (mostly from selectivity point of
view). Therefore, a short overview of the most important operation
modes is given below. For more details, please see Murray et al.
[61].
In SRM the precursor ion with previously specified m/z
(corresponding to the ion formed from the analyte) is fragmented
and the signals of one or more specific product ions are measured
[61]. This operation mode increases the selectivity of MS by
decreasing the probability of false positive identification. Ions
formed from matrix compounds with the same m/z as the analyte
ions generally have different molecular structures and thus yield
different product ions and will therefore not be detected. The
selectivity and reliability can be further improved if multiple
product ions from the same precursor ion are measured.
In MRM the signals of product ions from multiple precursor ions
are measured and used for quantitation of multiple analytes [61].
reason of occurrence of ionization suppression [41]. In case of a
perfect sample preparation combined with the perfect
chromatographic separation – leading to the chromatogram
where the analyte is completely separated from all matrix
components – ionization suppression/enhancement would not
occur and would not have to be considered. Unfortunately,
perfect sample preparation methods are not available in most
cases. A number of literature sources address choosing the
most effective sample preparation method from the matrix
effect point of view [47,48]. In LC–MS solid phase extraction
(SPE), liquid–liquid extraction (LLE), precipitation/centrifugation or combinations of these as well as other methods are used
for sample preparation. Bonfiglio et al. [47] compared different
sample preparation techniques and found that for phenacetin
and caffeine determination in endogenous plasma protein
precipitation is the least favorable technique for LC–ESI–MS
analyses while LLE was the most favorable. Additionally,
Souverain et al. [48] found LLE to be more effective sample
preparation technique than SPE for methadone determination,
because the latter tends to concentrate not only the analyte but
also matrix compounds similar to the analyte (i.e., potentially
co-eluting from HPLC with the analyte). The reason probably is
that for LLE a larger selection of extracting solvents is available
and therefore more freedom in varying selectivity is achievable.
On the other hand, in the case of SPE, a solid phase similar to the
HPLC stationary phase is often used (often both are low polarity
C18 or C8 phases) and therefore little additional selectivity is
obtained. Additionally, it has been shown by Dams et al. [49]
that sample preconcentration may significantly increase
ionization suppression.
(2) The main instrumental modification considered is using a nonESI ion source, such as APCI instead of ESI, since ionization in
the APCI source has been demonstrated to be less affected by
matrix effects [49–52]. Switching the ESI source from positive
to negative ionization mode [53] or reducing the flow rate of
the effluent [54], have also been demonstrated to be efficient in
Sometimes it is too difficult (and therefore impractical) or
impossible to remove all of the matrix effect, therefore, approaches
accounting for the matrix effect have also been developed. Most of
them fall either into the category of internal standard usage or
matrix-matched calibration and are described in the Part II of this
tutorial review [29]. Also some additional more complex
approaches have been proposed [44,59,60] in the literature but
are rarely used in routine analysis. Therefore, these methods are
not considered further in this paper.
2.3. Operation modes of MS as a detector
34
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
Consequent reaction monitoring (MSn) is similar to SRM, but
involves more fragmentation steps: the product ions of one
fragmentation are used as precursor ions for a next fragmentation
and so on, until the ions are eventually detected. n denotes the
number of fragmentations plus one. In this context, SRM can be
called MS2. NLS is monitoring the loss of an uncharged species from
precursor ions [61]. In this text, the general acronym MS (without a
superscript) means mass spectrometry with or without fragmentation.
2.4. Practical aspects of LC–MS method development
LC–MS offers a large variety of ionization methods, MS
operation modes and quantitation methods. This tremendously
increases the opportunities but also the responsibility of the user.
When developing an LC–MS method some important aspects to
consider are:
(1) There are a very large number of parameters in LC–MS methods
that can be optimized.
(2) Some of the key performance characteristics – most notably the
efficiency of ionization of the analyte in the ion source – are
either difficult to control or are very sensitive to small changes
in system parameters.
(3) MS detector in general displays inferior repeatability compared
to most other detectors (in their respective working concentration ranges), especially the UV–vis absorbance detector. The
repeatability standard deviation of MS signal, even if replicate
samples are analysed within a short time period, can be quite
high. The trueness of the quantitative MS results is decreased
first of all by different ion source related phenomena, such as
ionization suppression/enhancement. This has important
implications for determining trueness, precision and accuracy.
(4) MS as a detector is mostly used for determination of very small
quantities of analyte. Therefore, a number of problems, such as
incomplete selectivity, non-ideal sample preparation, etc. can
be even further amplified.
The benefits of MS as a detector (selectivity, identification
ability) have led to very high expectations by LC–MS users which
often do not become true. A most dangerous of them is the belief
that the importance of chromatographic separation as well as
sample cleanup becomes less important because the MS detector
provides selectivity of its own [37,41] and consequently very short
LC columns can be used. Unfortunately, even though the right
choice of MS operation mode often guarantees sufficient detection
and identification ability, the accuracy of the obtained results may
be strongly influenced by the ionization suppression/enhancement
occurring in most API sources. Both ionization suppression [41] or
ion source contamination [62,63] result in variability of both MS
signal and the obtained results.
and the International Union of Pure and Applied Chemists (IUPAC)
[16] have developed guidelines for single laboratory method
validation. The United States Food and Drug Administration (FDA)
has a guidance document for analytical method validation for
bioanalytical methods [18]. Similarly, the European Medicines
Agency (EMA, formerly EMEA) has a “Guideline on bioanalytical
method validation” [19]. The general Eurachem guide “The Fitness
for Purpose of Analytical Methods” [15] and NordVal “Guide in
Validation of Alternative Proprietary Chemical Methods” [26] are
not limited to any technique and can be used throughout the field
of analytical chemistry. There are also more specialized guidelines,
such as validation guideline for pesticide residue analysis in food
and feed by the SANCO [65] or European Commission Decision
2002/657/EC [66], that establish criteria and procedures for the
validation of analytical methods to ensure the quality and
comparability of analytical results generated by official control
laboratories.
Depending on the field of use, different guidelines may vary in
depth and also in parameters described. In general, most guidelines are targeted towards quantitative measurements. Even
though qualitative procedures are part of some of the guidelines
[20], these are not exhaustively discussed in this review.
In addition to previously mentioned official guidance documents, a number of articles, including review articles, have been
published on the topic of analytical method validation [6,21,64].
Most of these are general but some address specific applications or
techniques. There have been number of articles specifically
focusing on the LC–MS method validation in toxicology
[25,64,67]. In addition, there is a large number of articles
demonstrating validation of LC–ESI–MS methods for practical
analytical tasks with respect to certain validation guidelines such
as FDA [68–72], ICH [73–75] Eurachem [76–83], EMA [84–86],
IUPAC [87–89] or SANCO [90,91]. However, in most cases not all
parameters are investigated. There are also a number of cases
when for newly developed methods thorough validation is carried
out but not according to any of the published guidelines [92–95]. In
a number of cases only selected validation parameters are under
close interest and most commonly these are ionization suppression/enhancement, limits of detection, precision parameters
[85,96] and in many cases also recovery [97,98].
Many guidelines try to harmonize definitions required for
validation characteristics and their basic requirements, especially
the ICH guide, but they provide only a basis for general discussion
of the validation parameters, their calculation and interpretation.
Therefore, a lot is put on the analysts’ shoulders. They have to (1)
identify parameters that are relevant to the performance of the
given analytical method, (2) design of a proper validation protocol
including acceptance criteria and (3) choose the methods for the
appropriate data evaluation [99].
4. Parameters of LC–MS methods
3. Overview of the main validation guidelines and general
situation with validation of LC–MS methods
In the following sections the main performance characteristics
of analytical methods are examined with emphasis on LC–MS.
Owing to the importance of method validation in the whole
field of analytical chemistry, various international organizations
and conferences have issued a number of guidance documents
targeted toward single laboratory validation. Though different in
suggestions and requirements, all of these documents are
important and potentially helpful for validating any method [64].
Harmonized guidelines between European Union (EU), Japan
and the United States have been developed within the expert
working group of the International Conference on Harmonization
(ICH) of technical requirements for registration of pharmaceuticals
for human use (ICH Q2(R1]) [17] AOAC International (AOAC) [20]
4.1. Selectivity, specificity, confirmation of identity
Selectivity and specificity refer to the ability of a method to
measure the amount of the analyte that is claimed to be measured.
IUPAC suggests using term “selectivity” to express the extent to
which other substances interfere with the determination of an
analyte, while “specificity” is used to denote the ultimate
selectivity, meaning that no detectable interferences are supposed
to occur [100,101]. Analytical techniques are almost never specific
by themselves, but validated analytical methods within their scope
of application can be specific. Sometimes the term “specificity” is
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
used to denote selectivity. For example, ICH and NordVal use the
term “specificity”, while AOAC, EMA, Eurachem, FDA and IUPAC use
“selectivity”. In this paper we use the IUPAC recommendation.
Selectivity of a method has to guarantee that the effects of
interferents (degradation products, metabolites, etc.) in the
analyte signal are insignificant. The interferent can either act as
the analyte, and yield the signal indistinguishable from analyte’s
signal, or it can suppress/enhance the signal by altering sample
preparation, chromatographic separation or detector response. In
this chapter the former is dealt with, while the latter is discussed in
the Part II of this article [29].
Samples almost always contain some compounds that have
properties similar to the analyte. Therefore, the identity of a signal
presumably belonging to the analyte has to be confirmed.
Validation has to demonstrate that it is possible to confirm the
identity of the analyte, but identity confirmation in the actual
samples has to be done during routine use of the method. LC–MS is
often used for determining banned substances. In such analysis the
large majority of samples are “negative” (analyte missing). In such
cases it is sometimes useful to have two separate methods –
screening and confirmatory methods [20,65,66]. A screening
method is simple, fast, can be qualitative and is deliberately
positively biased (i.e., the probability of false negatives is
minimized). For example, for analysis of 120 pesticides and
metabolites in infant food Anagnostopoulos et al. developed an
MS2 screening method with one transition for each analyte and
separate confirmatory methods for each analyte containing two
transitions [102]. For the analysis of emerging environmental
contaminants [103] (e.g., residues of personal care products and
pharmaceuticals) qualitative screening methods based on liquid
chromatography coupled to high resolution mass spectrometer
(LC–HRMS) have proven to be especially suited [104,105]. Both
screening and confirmatory methods have to be validated.
Validation of screening methods is not within the scope of this
review and interested reader is directed to the following works:
[104–107]. Clearly, analyzing selected samples twice is disadvantageous with respect to workload and methods, which are able to
screen and confirm in one run are most welcome. The MS
operation modes described above, make LC–MS one of the most
selective techniques available, which also enables identity
confirmation.
Different validation guidelines approach selectivity and identity confirmation at different level of detail. The majority of the
guidelines [16–18,26] are very general and applicable to any
analytical technique. Such guidelines do not specifically address
MS. In contrast the European Commission decision 2002/657/EC
[66] and the SANCO guide [65] discuss LC–MS extensively.
4.1.1. Selectivity (specificity)
Selectivity evaluation in LC–MS usually starts from chromatographic resolution (Rs) between the analyte and the closest
eluting peak. Rs at least 1.5 is required by AOAC [20] and 2 by FDA
[18]. The Eurachem guide requires demonstration of separation
from other components also on a column of different chemistry
[15]. In addition to this, AOAC, referring to FDA validation
guidelines, demands that no other compound should be detectable
when other selective methods are used – IR, NMR or MS [20].
Similarly, ICH finds that if one is unable to demonstrate that a
single analytical method is specific, two or more analytical
methods should be used [17].
These requirements of ICH and AOAC are really stringent, as
they neglect detector-side selectivity. For example, if there is a coeluting compound at the analyte retention time, then it might be
possible to choose detector or detector setting (e.g., UV–vis
wavelength, MS mode or MRM transition) such as it responds only
to the analyte. From the perspective of analyte detection, this kind
35
of combined “LC + detector” selectivity should be acceptable
(considering of course the possible ionization suppression, see
above), but it would not be acceptable by the rules of ICH or AOAC
as virtually all the organic compounds are detectable by infrared
spectroscopy (IR) and nuclear magnetic resonance (NMR). Out of
the validation guidelines studied, only Eurachem foresees the
possibility that interferents are not separated [15].
In different situations different requirements apply for the
selectivity achieved with chromatographic separation. For example, in the case of pharmaceutical preparations, all components of
the sample must be revealed. So, all the components are
considered analytes and must be selectively detected. Situation
is different if for example, pesticide residues are determined in
vegetables. Vegetable extract matrix is rich in compounds, which
may co-elute with analytes, and it would be advantageous if
detector would be insensitive toward those endogenous compounds, enabling selective detection of the pesticides.
For selectivity assessment, NordVal suggests analysis of a blank
sample and spiked blank samples [26]. Although it is suggested to
analyze a more concentrated extract of the blank to demonstrate
absence of signal at analyte retention time, NordVal does not
specify any criteria for acceptable signal from the blank. Similarly
to NordVal EMA and FDA require analysis of blank matrices (at least
from six independent sources), but they also set a clear limit –
signal from the blank at analyte retention time must be less than
20% of the limit of quantitation (LoQ) for the analyte [18,19]. The
FDA guide adds that if LC–MS is used, analysis of six blank matrices
may not be important, but matrix effects should be investigated
[18].
According to the European Commission decision 2002/657/EC,
specificity is evaluated by analyzing at least 20 representative
blank matrices and additionally blank matrices fortified with
compounds likely to interfere with the analysis. It should be
demonstrated that analyte is not falsely identified, its identification is not hindered and its quantification is not influenced [66].
While the Eurachem guide just notes that interferents usually
affect the slope of the calibration curve [15], IUPAC recommends
quantitation of selectivity by means of selectivity index, which is
defined as ban/bint, where ban is the slope of the calibration function
for the analyte and bint is the slope for the potential interferent
[16]. bint can be determined by analyzing blank sample and blank
sample spiked with the interferent. Availability of the interfering
compound limits the use of selectivity index. In addition, no
acceptance criteria are given in the guideline for the selectivity
index. The present authors are not aware of any literature source
demonstrating the use of the selectivity index.
In conclusion, although MS is a highly selective detector,
interferences are not uncommon in LC–MS analysis [108,109].
During validation the analyst has to test as many possible
interferents as reasonable, and continue monitoring selectivity
during routine use of the method. This can be done e.g., by
routinely analyzing a suitable control sample (either natural or
self-prepared) containing the analyte (a known amount) and
interferent(s).
4.1.2. Identity confirmation
For identity confirmation some of the validation guidelines
recognize the power of MS coupled to chromatography. AOAC
recommends LC–MS (or gas chromatography–mass spectrometry
(GC–MS)) full scan mass spectrum or identification of 3–4 fragments or MSn [20]. Eurachem states that if the method under
evaluation is not specific, confirmation by independent methods
must be performed [15]. FDA [18] and ICH [17] are rather general,
the former stating that “Evidence should be provided that the
substance quantified is the intended analyte” and the latter
requiring that positive results must be obtained by analyzing
36
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
samples containing the analyte and negative results from samples
not containing the analyte, and also negative results must be
obtained if structurally similar or closely related substances are
analyzed.
Perhaps the most concrete guidance about assessing identity
confirmation in LC–MS is given in the European Commission
decision 2002/657/EC [66] and the SANCO guide [65]. At least
7 data points must be acquired over a chromatographic peak [66]
and the shape of the analyte’s peak in the sample must be similar to
that obtained from the calibration solution [65]. The relative
retention time (i.e., the ratio of the retention time of analyte to that
of internal standard) should match that of the calibration solution
within 2.5% [65,66].
For identity confirmation 2002/657/EC [66] and the SANCO
guide [65] also set criteria for mass spectra – relative abundances
of analyte ions in sample compared to this in standard solution
should fall within specified ranges. For example, if the intensity of
an ion is within 10–20% relative to base peak, its maximum
permitted tolerance is 30% (for full list of criteria, see [65,66]).
Additionally it is suggested to compare the obtained mass spectra
to the reference spectra obtained with the same instrument.
For identification at least four ions with specified tolerances
[66] must be registered. Due to soft nature of ionization methods
used in LC–MS fragmentation of quasi-molecular ion is seldom
observed in single stage (LC–)MS. In addition, in triple quadrupole
instruments full scan spectra lead to significantly higher limits of
detection than monitoring specific ions. Therefore, fragmentationbased MS modes are recommended. The way of specifying the
amount of data required for identity confirmation is different in
2002/657/EC [66] and the SANCO guide [65]. European Commission [66] has introduced an elaborate system of identification
points, where, for example, application of MS/MS for isolating one
precursor ion (gives 1 point) and registering signals from two
fragments (1.5 points each) yields 4 identification points in total.
The required number of points for unambiguous identification is
3 or 4 depending on substance. The SANCO guide [65] simply states
that for identification using MS/MS experiment at least 2 product
ions must be analyzed – thus coming to the conclusion similar to
that of European Commission [66].
The system of identification points is often used in case of liquid
chromatography coupled to low resolution mass spectrometer
(LC–LRMS) analyses [52,110], but also in case of LC–HRMS methods
[73,111]. For the best identity confirmation the mass spectrometer
should be capable of precursor isolation-fragmentation and
registration of the resulting full mass spectrum at high resolution.
This is acknowledged by the European Commission decision [66]
by assigning to HRMS double the identification points of LRMS
(giving 2 points for a precursor ion or 2.5 for a transition). Utility of
HRMS for identification is also demonstrated by Pozo et al. [112].
The SANCO guide [65] requires registration of at least two
diagnostic ions (at least one fragment and preferably quasi
molecular ion). For the complete set of identification requirements
see Refs. [65,66].
Also, directions for choosing proper diagnostic ions, which
enable achieving selectivity, are given in documents [65,66].
Quasi-molecular ion, product (fragment) ions with higher m/z and
fragments that originate from different parts of the molecular ion
are recommended. Transitions associated with losses of common
fragments (for example loss of water), have been demonstrated to
cause false positive as well as false negative results [112] and
should therefore be avoided.
Criteria, similar to those of SANCO and European Commission,
have also been published by FDA [18] and Bethem et al. [113], but
are rarely referenced.
Compared to the validation guides (except SANCO), the
regulation 2002/657/EC has most thoroughly incorporated the
specific aspects of LC–MS. Therefore we recommend 2002/657/EC
as guidance for LC–MS method selectivity and especially identity
confirmation.
4.2. Ruggedness/robustness
4.2.1. Definitions
The terms robustness and ruggedness refer to the ability of an
analytical method to remain unaffected by small variations in
method parameters (mobile phase composition, column age,
column temperature, etc.) and influential environmental factors
(room temperature, air humidity, etc.) and characterize its
reliability during normal usage. The notion of remaining unaffected has two possible interpretations – it can be interpreted as (1) no
change of the detected amount of the analyte in a certain sample in
spite of the variation of the method parameter [66] or (2) no
change of the critical performance characteristics (e.g., limit of
quantitation) by the variation of the method parameter [65]. In
experimental evaluation of robustness either one of these
interpretations can be used.
The definitions in the guidelines [15–17,20] as well as review
articles [21,27,114,115] are very similar. Some guidelines use the
term robustness and some use ruggedness. When used together
they are treated as synonyms in most cases [7,15,66,114–120]. The
only widespread guideline making difference between these terms
is the former USP [121], but the later versions of USP [122] use only
the term robustness. Considering robustness as a method
development parameter, EMA and FDA guidelines are missing
the term. A recent review article emphasizes the importance of
robustness testing and discusses different approaches thoroughly
[27].
The above definitions imply changes made to the method
within the same laboratory. However, robustness can also be
described as the feasibility to reproduce the analytical method in
different laboratories or under different circumstances without the
occurrence of unexpected differences in the obtained results [118].
Along the similar lines it has been suggested that ruggedness
should be used as a parameter evaluating constancy of the results
when external factors such as analyst, laboratory, instrument,
reagents and days are varied and robustness should be used as a
parameter characterizing the stability of the method with respect
to variations of the internal factors (parameters) of the method
(e.g., parameters related to sample preparation, mobile phase
composition, mobile phase flow rate, injection volume, column
temperature etc.) [14,21,123].
The term robustness is in most cases understood in terms of
influence of variations of method parameters on results. Our
experience suggests, however, that an additional dimension –
robustness in terms of variability of sample matrix – is beneficial:
different matrices can lead to different matrix effects (either in the
narrow or broad sense). On the example of blood plasma:
depending on the personal variations in metabolism, diet, possible
diseases, e.g., the composition (first of all but not limited to the
content of proteins, phospholipids and polyunsaturated fatty acids
[124–126]) of blood plasma can vary significantly, even though
formally the matrix is the same – blood plasma [127]. The sample
preparation procedure that is suitable for blood plasma of low
protein or phospholipid content may give different results for
blood plasma with high protein or phospholipid content. This is
occasionally addressed in validation guidelines under selectivity
[18,19]. However, the possible effects of this kind of variability are
not limited to the loss of selectivity, but can also influence recovery
(and hence trueness), ionization suppression/enhancement as well
as limit of detection (LoD)/limit of quantitation (LoQ). It is thus
useful to investigate the effect of sample matrix variability (in the
case of formally identical matrices) more broadly than just for
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
selectivity. See also the section of Trueness in part II of this review
[29].
There are studies where the terms robustness/ruggedness are
misinterpreted and actually decision threshold, detection capability (see below) [128] or measurement uncertainty [88] is
evaluated.
In this review we use the term robustness for expressing the
stability of the method against small variations of the intrinsic
method parameters. Since some changes in the method performance can occur over longer period of time, the robustness is a
validation parameter that has to be monitored during the
validation and also after the validation procedure and during
the method lifetime. We use the term ruggedness for expressing
the stability of the method against extraneous influencing factors.
We address changes of the method parameters (i.e., withinlaboratory assessment of robustness) and variability of sample
matrices. We do not explicitly address changes occurring when a
method is transferred from one laboratory to another.
4.2.2. Specific aspects in LC–MS
As explained above, there are a very large number of adjustable
parameters in LC–MS methods and some of the key performance
characteristics – most notably the ionization efficiency of the
analyte in the ion source – are either difficult to control or are
sensitive to small changes in system parameters (or sample
properties), resulting in poor instrument reproducibility between
the runs (between samples with formally identical matrix).
Furthermore, LC–MS is very often used for the determination of
very low levels of analytes and in highly complex matrices.
Analysis of complex matrices often requires complex multi-step
sample preparation procedures.
Put together, the above listed factors clearly indicate that in the
case of LC–MS method validation, investigation of ruggedness and
robustness is very important.
4.2.3. Experimental design of ruggedness/robustness testing
Because of the very large number of potentially variable
parameters it is reasonable to divide assessment of ruggedness into
separate parts. A very logical division would be to test ruggedness
separately for sample preparation and for the LC–MS analytical
part.
The experimental designs suggested for validation often involve
fractional factorial (saturated designs) [129] or the Plackett–
Burmann design [119,130,131].
In the Plackett–Burmann approach, N 1 parameters (variables) are studied in N experimental runs (with N being a multiple of
4) [22,118,130]. The factors are investigated at two levels: low ( )
and high (+). From the obtained signals, factor effects are calculated
according to:
P
P
Yð Þ
YðþÞ
;
Ex ¼
(1)
N=2
where Ex is the effect of variation of parameter X, the sums indicate
the signals where factor X is at (+) or ( ) level. N is the number of
design experiments. Both graphical and statistical interpretation
methods are described [22,129,132,133] for identifying significant
effects.
Parameters X indicate the different factors (operational factors
and environmental factors) and their levels on robustness
estimation. The factors can be divided into quantitative (pH,
temperature, concentration of the solutions), qualitative (column
age, recent changes of chromatographic column made by the
manufacturer) and mixture factors (fraction of the organic
modifier on the eluent) [118].
In the saturated designs, the number of experiments required,
without counting the central (initial) points, is equal to the number
37
of variables plus one. However, none of the main effects are
confounded with each other. If the number of variables is large
(e.g., MS parameters are also included) and the number of factors
examined exceeds the number of experiments, the supersaturated
designs are used [118]. The total variance of changes in results
could be used as a measure of robustness. The significance of the
effects of the parameters (factors) on the signals can be evaluated
using ANOVA or t-test (in case the relationship between the
parameter change and signal is linear) [27,119].
The above described approaches are rigorous and powerful.
However, literature survey, as well as contacts with routine
laboratories indicate, that these approaches are seldom used
[92,134–138]. The main reason is that these approaches require
knowledge and experience with mathematical statistics. In most
cases experiments with one-by-one variations (one variable at a
time approach) of the most important parameters are carried out
[27,138–142].
Concluding, there is a difficult situation where on the one hand
analytical methods are very complex, have numerous parameters
and would, thus, benefit from validation with full rigor and, on the
other hand, available resources are always limited. Before starting
investigation of robustness it is crucial to find out, what are the
critical performance characteristics of the method. According to
these characteristics the method parameters to be varied are
thereafter chosen. For example, if LoQ is very close to the LoQ
required by legislation, then changes in LoQ value have to be
monitored against small changes in method parameters. Most
influential method parameters for LoQ could be MS parameters,
mobile phase pH, extraction parameters (see Table 1). The main
criteria for choosing parameters are (1) how much a given method
parameter can influence the critical characteristic and (2) how
likely it is that this parameter will change uncontrollably.
Based on the common practice in literature and on our own
experience we recommend the following:
1. Change parameters one by one (one variable at a time approach)
in both directions from the nominal (optimal) value. Changes in
the parameters should be realistic in the context of normal use
of the method.
2. Do not stop there! Often parameters may be mutually unrelated
(uncorrelated), but in some cases this does not hold. For
example, change in mobile phase pH can decrease resolution
between two adjacent peaks. Likewise, increase of mobile phase
flow rate can also lead to decrease of resolution. While
separately any of these two changes can still lead to no loss
of resolution, their occurrence together may lead to peak
overlap. Whether this is the case, can often be determined by
educated inspection of the effects of the changes (without
additional experiments) and noting possible problems.
3. Effects from the change of parameters should be monitored. If
necessary, graphical or statistical analysis of the effects should
be done.
4. Regarding the robustness tests results, if necessary, measures to
improve the performance of the method should be taken.
4.2.4. Method parameters to be varied
The validation guidelines differ somewhat by their suggestions
on which parameters should be varied and by how much.
ICH [17] states different conditions such as variations in mobile
phase composition, usage of different columns (different ages,
different lots), column temperature and flow rate.
Eurachem [15] recommends using the AOAC guide and Youden
ruggedness trial giving a short overview of ruggedness tests and
the influence on either precision or accuracy. Different parameters
that could be varied and taken into account are for example,
38
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
Table 1
Recommended method parameters to be investigated during robustness studies.
Parameter
Likelihood of uncontrollable
change
Recommended extent of variation
Comments
Medium
0.5 units
Concentration of additives in
eluent
Medium
10% (relative)
Organic solvent content in the
eluent
Column temperature
Low to Medium
2%
Low
5 C
Eluent flow rate
Low
20%
Column batch and age
Medium
–
pH will have a strong effect on retention time (and
possibly resolution) if the analyte’s pKa value is within
1.5 units of the mobile phase pH
Salts, ion-pair reagents, modifiers can suppress/
enhance analyte’s ionization in the ion source and
change its retention time and possibly resolution from
other compounds
Organic solvent content influences retention time (and
possibly resolution) and analyte signal in LC–MS
Column temperature influences the retention time
(and possibly resolution)
Eluent flow rate influences the retention time (and
possibly resolution)
Changes in column can influence the retention time
(and possibly resolution)
Liquid chromatography
pH
Samples and sample preparation
Analyte extraction time; solvent
High
amount and composition (in
liquid/liquid and solid phase
extraction, etc.)
Injection solvent composition
Low/High
Matrix effect in broad sense
(sample matrix source)
Mass spectrometry
Drying gas temp
20%
Influences recovery and LoQ/LoD
10% (relative)
This is the solvent in which analyte is taken up during
the last stage of sample preparation. The
recommended extent of variation refers to the minority
component(s). This composition can influence
retention time and recovery and therefore also the
matrix effect (in broad sense), LoQ/LoD and stability.
The effect can be very serious in the case of UHPLC and
is usually not that critical in the conventional HPLC
Can be assessed under selectivity studies [19,18].
Influences trueness (recovery and ionization
suppression), LoQ/LoD
High
6 different
Low
10 C
Nebulizer gas pressure/flow rate
Low
Ion source configuration
(nebulizer position)
High (if configurations can vary) According to the ion source design
Not applicable (if fixed source)
Ion source condition (nebulizer
High
aging, ion source contamination)
5 psi/ 1 L min
1
Should be varied if source is used in
different configurations
After analysis of samples versus
cleaned system
different analysts, instruments, reagents and variations in sample
preparation or sample matrix. Evaluation of each parameter
separately gives useful information about method robustness.
From the results the effect of changes on method can be estimated
and the factors can be determined for the parameters [143].
The NordVal guide [26] lists the most popular parameters to be
included in the robustness test: the composition of the samples,
mobile phase pH, timing of individual (assay) steps, temperature
and presence of potentially interfering substances (e.g., tannins or
other complexing agents, varying levels of lipids, endogenous
enzymes).
Robustness can be studied by varying typical parameters in the
chromatographic run that are capable of influencing analysis
results [144] as well as the influence of the columns age or
replacement. Our suggestions are presented in Table 1. It is well
known that even if two columns are formally of the same type their
performance can differ [145]. Additionally, it may be necessary to
investigate robustness related to change of the instrument or some
of its components. This need will occur when either a component
(e.g., pump or detector) or the whole system is replaced.
In the literature limited attention has been paid to varying
besides the LC parameters also the MS parameters [146,147]. Our
Drying gas temperature can influence analyte
ionization efficiency in the ion source
Nebulizer gas pressure/flow rate can influence analyte
ionization efficiency in the ion source
Ion source configuration can influence spray and
ionization efficiency in the ion source
Contamination can spontaneously accumulate when
analyzing a series of samples
experience shows that in the case of LC–MS methods the MS
parameters are as important as the LC parameters (although
several problems, such as matrix effects due to overlapping peaks,
analyte peaks falling outside the MS time window, etc are caused
first of all by LC).
Our recommendations on the choice of method parameters to
be varied during robustness study are collected in Table 1.
Additionally, the possible changes of the critical performance
characteristics (e.g., LoQ) and detected amount of analyte (e.g.,
decrease of trueness because of interference caused by incomplete
resolution) are commented on. Differently from the LC–MS
method itself, sample preparation can have a large number of
parameters that cannot be easily systematized. Therefore the
recommendations about sample preparation parameters are very
general.
4.2.5. Numerically expressing robustness/ruggedness
Robustness/ruggedness is usually expressed as relative standard deviation (RSD)% of data obtained with the changed
parameter inside the method with respect to the same data
obtained using initially observed conditions. Expressing the
robustness using the RSD [133] is a simple and well understood
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
approach. There is some similarity between RSD expressing
robustness and RSD expressing intermediate precision (see Part
II of the review [29]). The main difference is that in the case of
robustness determination the changes in the method are introduced deliberately and their magnitude is controlled, while
intermediate precision addresses unintentional changes that occur
under normal operation of the method in the laboratory. In
addition, robustness/ruggedness can be evaluated by the Student ttest to assure the statistical significance of these obtained results
[144]. The review article [133] focuses on the variety of statistical
tests (e.g., the t- and F-tests) for the robustness evaluation. These
test the statistical significance of the changes obtained by the
robustness evaluation. In addition, it has to be carefully evaluated if
the changes are also significant from the fitness for purpose
perspective, compared to other parameters influencing the
method performance.
4.3. Limit of detection, limit of quantitation
4.3.1. Limit of detection
Limit of detection (LoD), also called detection limit, is loosely
defined as the lowest amount or lowest concentration of the
analyte in a sample which can be reliably detected and identified
with the method [15–17,26]. This does not imply the possibility to
quantify the result [26]. Although this definition is used in many
cases due to its simplicity it is not fully rigorous, because the
meaning of “reliably” is not clarified. Furthermore, when speaking
about detecting an analyte near the detection limit, two types of
false results are possible – false positive and false negative results –
which are not fully accounted for in this definition. The common
way of interpreting LoD data takes care of avoiding false positives,
but tolerates false negatives: If a peak that could belong to the
analyte is there but the calculation gives a value below LoD then it
is not possible to reliably declare that the analyte has been
detected (and the result has to be reported as “below LoD”),
because in reality the analyte may be present but at a very low
concentration. Interestingly, explanation of this meaning of LoD is
not included in validation guidelines.
An alternative approach of characterizing the detection
capability of a method has been proposed to account for these
shortcomings. This is achieved by specifying the lowest concentration levels related to analyte detection reliabilities in terms of
probabilities of false positive and false negative results. These
lowest levels are termed as decision limit (CCa) and detection
capability (CCb), respectively. This approach was first proposed by
Currie in 1968 [148], is briefly mentioned in the IUPAC validation
guideline [16] and more recently adopted into the EU food safety
legislation [66]. CCa and CCb are discussed further in a following
chapter.
We discuss here determining the LoD of the “whole method”,
expressed as the analyte concentration in the sample, e.g., mg of
analyte per kg of sample. This means that LoD should be
determined taking into account all steps in the method
(including sample preparation). We do not discuss here the
so-called instrument LoD (which is related to detecting analyte in
pure solvent). Since any alterations made to the method may
change LoD, LoD should be determined for fully developed
methods [16]. For analytical methods where LoD is not in the
validation range it is only necessary to estimate that it is low
enough, but its numerical determination is not necessary [16].
LoD is a highly variable parameter (i.e., it has poor day-to-day
reproducibility) and it should therefore be re-determined
regularly in order to reflect the actual operating performance
[15]. Therefore, a conservative estimate of LoD is preferred,
especially for cases where LoD is calculated from the data
collected during a short time period. If LoD is estimated on more
39
than one day, the highest LoD value (i.e., the most conservative
LoD value) should be used so that the declared LoD would be
routinely achievable with high probability. It is important to
stress that the word “conservative” here means that the
probability of false positives will become lower, while the
probability of false negatives will become higher.
LoD is matrix-sensitive and should be determined in a matrix
that matches the real sample matrix [15], i.e., using matrixmatched samples. It can be difficult or even impossible to find
blank matrix-matching samples and therefore a blank sample with
a similar matrix can be used (see Chapter 3.2.2 in Part II of this
tutorial review [29]).
In the case of LC-MS/MS, multiple product ion intensities are
often monitored to confirm the analyte identity. The European
Commission decision 2002/657/EC and SANCO set the requirement
that S/N ratios for all diagnostic ions should be 3. The same
guidelines also set the maximum allowed errors of relative ion
intensities and the number of diagnostic ions necessary for
different instrumentation and chemicals [65,66]. These requirements markedly influence LoD determination: LoD has to address
the diagnostic ion with the lowest intensity because as soon as the
presence of this ion cannot be stated with confidence, the identity
of the contaminant becomes uncertain. Different approaches for
determining LoD are discussed below.
Signal-to-noise ratio (S/N) can be used to determine LoD if the
method exhibits baseline noise. S/N is found by comparing signals
from samples with known low analyte concentration and blank
samples. LoD is determined by establishing the minimum
concentration at which S/N is over 3 or between 3 and 2 [17].
This approach cannot be always used with LC-MS/MS measurements, because the baselines in the MS/MS mode do not always
exhibit noise. When the noise level cannot be reliably measured
(either there is no baseline noise or it is irregular) then the S/N
values become highly variable or S/N cannot be calculated at all.
However, if noise can be measured then the data analysis
programs usually enable calculation of S/N that can also be used.
It is recommended to use the latter method to find S/N as it
provides values with lower variability. The S/N ratio can vary
significantly between days and even within a day, thus many
measurements should be made for reliable determination of LoD
[149].
Another approach to determine LoD is to measure a number of
solutions with different concentrations close to the LoD, each with
10 separate samples. The lowest concentration where all 10 samples have positive results can be considered LoD [15,26]. This
approach was originally designed for qualitative analysis where
the result is expressed either as a positive (analyte detected) or a
negative (analyte not detected). Nevertheless, it can also be used
for quantitative analytical methods. This is so because, in fact, in
the LoD region most quantitative methods become qualitative. The
difference between positive and negative samples can be made for
example, based on the S/N: a concentration level where all
10 samples have S/N over 3 can be used as LoD. Alternatively, a
visual assessment of the LoD can be used.
If a blank sample can be measured, the following equations can
be used to determine LoD:
LoD ¼ 3 sðblankÞ
(2)
or
LoD ¼ X ðblankÞ þ 3 sðblankÞ
(3)
where X(blank) is mean value of the blank results and s(blank) is
the standard deviation of the blank values [15,26]. However,
Eurachem requires the use of fortified samples for chromatographic techniques as measurement results can be obtained only when a
peak can be detected over noise [15]. The following equations can
40
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
be used for the fortified samples:
LoD ¼ 3 sðfortifiedÞ
(4)
or
LoD ¼ XðblankÞ þ 4:65 sðfortifiedÞ
(5)
where s(fortified) is the standard deviation of the fortified sample
[15,26]. Samples should be fortified at the concentration level close
to the LoD or at lowest acceptable level determined by the needs of
the method [15]. Eurachem also suggest an approach to take into
account the number of repeated sample measurements made by
averaging the repeated results [15]. In case blank correlation is
used, Eurachem also suggests taking into account the repeated
measurement made for blank and the sample [15]. If the
calibration graph intercept is statistically insignificant then there
is no difference between using concentration scale (after
calculating concentration values for all intensities) or intensity
scale (after which LoD as concentration can be found from that
value) in the above equations. However, if the intercept is
statistically significant then only concentration scale can be used
[149]. The number of replicate determinations is important,
because unless a large amount of data is collected LoD will be
subject to large random variations [16]. Eurachem and IUPAC
recommend that at least 10 and 6, respectively, separate blank and
fortified samples should be measured [15,16]. If the intercept is not
significantly different from zero it can be left out of the
calculations. This is necessary because in case of very large
intercept values negative apparent values of LoD can be obtained,
which do not have physical meaning [149].
Another approach to determine LoD is from the equation
LoD ¼ 3:3 s
b
(6)
where b is calibration graph slope and s is the standard deviation of
the signal. s can be determined in 3 different ways. First, s can be
determined as standard deviation of an appropriate number of
blank sample signals. Second, s can be determined as residual
standard deviation of the calibration line in the LoD region. Third, s
can be determined as the standard deviation of intercept [17].
Visual evaluation can also be used to determine LoD. This is
done by performing analysis of samples with known analyte
concentrations and by visually establishing the minimum level at
which the analyte can be reliably detected [17]. Usually peak
shapes and heights vary between samples with the same
concentration. Therefore multiple measurements should be made.
Other approaches have been suggested in articles on LoD
determination. These, for example, use information theory to take
into account different distributions of noise in different analytical
systems [150] or suggest reporting LoD with confidence limits
[151]. However these approaches are not widespread or generally
accepted.
The multitude of possibilities of LoD determination can make it
difficult to choose the most suitable approach. In a recent study
[149] careful comparison of 10 LoD determination approaches was
carried out by some of the authors of this review for the LC-MS/MS
analysis of meropenem, doripenem and cilastatine. The LoD values
obtained with different approaches differed by up to an order of
magnitude. It was shown that comparison of LoD values found by
different approaches is not meaningful. Based on the study results
the recommended LoD determination approach is the one using
Eq. (6) where s is the residual standard deviation of the regression
line. This approach is recommended as it gives conservative LoD
estimates while using measurements made for calibration graph. It
is important that if samples with analyte content close to LoD are
suspected then calibration points in the range of LoD are included.
This approach of LoD determination requires only little extra work
compared to routine analysis and enables frequent LoD determination so that the determined values correspond to the actual
situation at the laboratory [149].
In conclusion, we advise determining LoD at least 5 times over a
long period (e.g., 5 months) to acquire a sufficiently representative
LoD result. The highest obtained LoD should be declared as the
limit for the method, so that the laboratory would be with high
probability routinely able to achieve this LoD. Although examples
can be found in the literature where LoD has been determined on
different days [87] in most cases LoD is determined only once
[128,138,152]. It is also recommended to visually evaluate the
peaks in the LoD range to be sure that the calculations have not
given unrealistic values. When stating LoD of the analytical
method, the approach used to determine LoD should be specified
[17]. If the result of the sample on a specific day is visually above
the LoD and a control sample is measured that shows lower LoD
than previously reported then the analyte can be reported as
present in the sample. Moreover, the performance of the method
should be monitored regularly by analyzing samples in the vicinity
of LoD or a little above. Problems with the measurement method
are indicated by significant variation of results from the average or
results drifting in one direction away from the average over a
longer period of time. In these cases the performance of the
method should be reevaluated and if necessary the source of
variance should be determined and fixed.
4.3.2. Limit of quantitation
Limit of quantitation (LoQ) is defined as the lowest concentration of analyte that can be determined with an acceptable
repeatability and trueness [15]. LoQ is called lower limit of
quantification [19], limit of quantitation [15], limit of quantification [16,18,26], quantification limit [153], quantitation limit [17] or
limit of determination [15,16,20] in different standards and
guidelines. As with LoD, this definition is also not fully rigorous,
but a more rigorous substitute, which would be widely accepted, is
not available.
Note that by definition quantitation is possible at either LoD or
LoQ but only the associated uncertainty becomes comparable to
the actual result when approaching LoD [15]. Therefore, LoQ is just
an indicative value as results below it are not devoid of information
and can be fit for purpose. In case the result of a measurement is
between LoD and LoQ it can be reported that the analyte presence
has been detected in the sample but is below LoQ [154]. Moreover,
it is suggested that LoQ should be found by expressing the
uncertainty of the measurement as a function of concentration and
comparing the results to the uncertainty levels demanded of that
method [15,16]. In this case at least 10 repeated measurements
should be made in each calibration point [15].
Approaches to determine LoD and LoQ are similar. In the latter
only a greater multiplication coefficient in the equation is used or
other higher demands are set on the same parameters. For
example, S/N value of at least 10 is required at the LoQ
concentration level [17]. When using the approach with equation
LoQ ¼ XðblankÞ þ k sðblankÞ
(7)
the same variables can be used and the same amount of analyzed
samples is required. However, the coefficient k is required to have
values of 5, 6 or 10 [15,18,19,20,26]. In case of the ICH approach,
using the calibration line, the equation
LoQ ¼ 10 s
b
(8)
is used where again all the variables can be taken from the same
datasets for both LoD and LoQ [17]. When using visual evaluation,
the LoQ is taken as the lowest concentration level where the
analyte can be quantified with acceptable level of precision and
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
trueness. [17]. It has also been suggested that LoQ can be found by
multiplying LoD by 2 [16].
Another approach to determine LoQ is to find precision,
trueness or recovery of the method at multiple concentration
levels. LoQ can then be taken as the lowest concentration where
these parameters are fit for purpose or meet the requirements of
the necessary legislation [15,18,65]. For example, FDA requires that
the intensities of the peaks at LoQ must have precision of 20% and
trueness of 80–120% [18], and SANCO requires that mean recovery
is in range of 70–120% and relative standard deviation of at least
20% [65].
LoQ is determined in most approaches from the same data as
LoD or is based on LoD and therefore in principle the same issues
occur. Nevertheless, as LoQ is higher than LoD and therefore the
results in the LoQ range have lower relative uncertainty, which
makes LoQ determination more reliable.
In our experience, the most appropriate approach to determine
LoQ is by using the Eq. (8) suggested by ICH where s is taken as
standard deviation of the calibration line residuals in the low
concentration range. Moreover, LoQ should be determined 5 times
over a longer period and the most conservative result should be
stated as the methods’ performance level to increase its reliability.
The exact way of determining LoQ should be specified as with LoD
due to the differences of the results when different approaches are
used. Moreover, the methods’ performance at the LoQ level can be
monitored with regular analysis of samples (either real contaminated samples or spiked blank samples) with concentrations close
to LoQ.
4.3.3. Decision limit (CCa) and detection capability (CCb)
In order to determine CCa and CCb a critical concentration
needs to be defined above which the sample is said to contain the
analyte and below which no analyte is said to be present. When a
permitted limit is not established for the analyte, the critical value
is defined as the concentration at which the probability of a blank
sample being considered as a positive sample (containing analyte)
is a. If a is 0.05 (this is the commonly used probability level) then
this means that the critical value is the concentration value above
which the results of 5% or less of the blank samples lie and are
falsely considered positive. This critical value is called CCa [66].
However, when the analyte level in the sample is indeed above CCa
then there is a danger of obtaining a result that is below CCa and is
therefore falsely counted as negative. This danger is not addressed
by CCa. Therefore, after finding CCa we can define an analyte
concentration at which the probability of getting a falsely negative
result is b. This concentration level is often termed as CCb. If b is
0.05 then CCb means a concentration where the analysis of the
sample would give results under the critical value only 5% of the
times. CCb can be interpreted as LoD and some guidelines define
LoD this way [153,155] others term this concentration as CCb [66].
In case a permitted limit for the analyte is established, the
critical value is found by analyzing blank samples that are
spiked at that permitted limit, not from the blank samples.
Therefore, if a is taken as 0.05, CCa is the concentration above
which 5% of the highest results lie (and are considered falsely
positive) that are obtained for samples with analyte concentration at (i.e., not exceeding) the permitted limit. CCb is then the
concentration at which the probability of obtaining results under
the respective CCa (i.e., obtaining false negatives) is 5% (if b is
taken as 0.05) [66]. The explanation of CCa and CCb can be found
in Fig. 1.
By many standards and guidelines calculating CCa and CCb
needs complex understanding of the statistics behind these
definitions [155]. However simpler methods exist that are
presented here. CCa can be defined as
41
Fig. 1. Relation between permitted limit, CCa and CCb [66].
CCa ¼ X þ k sR
(9)
where X is intercept of the calibration graph, k is 1.64 or 2.33,
depending on whether a is taken as 0.05 or 0.01 respectively, and
sR is the between-day reproducibility of blank samples or at the
permitted limit set for that analyte [66]. From here CCb can be
found as
CCb ¼ CCa þ k sR
(10)
where k and sR are the same as for CCa calculations. At least 20
replicate samples should be measured to calculate sR [66]. These
calculations do not take into account the increase of standard
deviation with the increase of concentration.
As more replicate measurements and more specific definitions
are used here compared to the LoD determination approaches, the
results are more reliable. Therefore CCa and CCb should be used
when the analysis method is operating at analyte concentrations
close to LoD and reliable decisions, whether the analyte is present
or not, are necessary. Although this approach increases the time
needed to determine LoD due to its superior reliability it is widely
used in method validation [110,156,157].
Acknowledgments
This work was supported by the institutional research funding
IUT20-14 and personal research funding PUT34 from the Estonian
Ministry of Education and Research, as well as the Estonian Science
Foundation grant No 8572 and project “Development of software
for validation of chromatographic methods” (with registration
number 3.2.1201.13-0020) under the sub-measure “Supporting the
development of R&D of info and communication technology”
funded by the EU Regional Development Fund.
References
[1] J.H. Gross, Mass Spectrometry, second ed., Springer, Heidelberg, 2011.
[2] W.A. Korfmacher, Foundation Review: principles and applications of LC-MS
in new drug discovery, Drug Discovery Today 10 (2005) 1357–1367.
[3] B.W. Brooks, D.B. Huggett, Human Pharmaceuticals in the Environment.
Current and Future Perspectives, Springer, New York, 2012.
[4] A. Kruve, K. Herodes, I. Leito, Optimization of electrospray interface and
quadrupole ion trap mass spectrometer parameters in pesticide liquid
chromatography/electrospray ionization mass spectrometry analysis, Rapid
Commun. Mass Spectrom. 25 (2010) 1159–1168.
[5] M. Jemal, Y.Q. Xia, LC–MS development strategies for quantitative
bioanalysis, Curr. Drug Metab. 7 (2006) 491–502.
42
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
[6] I. Taverniers, M. De Loose, E. Van Bockstaele, Trends in quality in the
analytical laboratory. II. Analytical method validation and quality assurance,
TrAC Trends Anal. Chem. 23 (2004) 535–552.
[7] E. Trullols, I. Ruisánchez, F.X. Rius, Validation of qualitative analytical
methods, TrAC Trends Anal. Chem. 23 (2004) 137–145.
[8] I.R. Pizzutti, A. de Kok, R. Zanella, M.B. Adaime, M. Hiemstra, C. Wickert, O.D.
Prestes, Method validation for the analysis of 169 pesticides in soya grain
without clean up, by liquid chromatography–tandem mass spectrometry
using positive and negative electrospray ionization, J. Chromatogr. A 1142
(2007) 123–136.
[9] A. Van Eeckhaut, K. Lanckmans, S. Sarre, I. Smolders, Y. Michotte, Validation of
bioanalytical LC-MS/MS assays: evaluation of matrix effects, J. Chromatogr. B
877 (2009) 2198–2207.
[10] E. Rogatsky, D. Stein, Evaluation of matrix effect and chromatography
efficiency: new parameters for validation of method development, J. Am. Soc.
Mass Spectrom. 16 (2005) 1757–1759.
[11] ISO/IEC 17025,2005, General requirements for the competence of testing and
calibration laboratories.
[12] GLP–OECD Series on Principles of Good Laboratory Practice (GLP) and
Compliance Monitoring.
[13] Commission directive 2009/90/EC: Laying down, pursuant to Directive 2000/
60/EC of the European Parliament and of the Council, technical specifications
for chemical analysis and monitoring of water status. Off. J. Eur. Union, 2009.
[14] P. Araujo, Key aspects of analytical method validation and linearity
evaluation, J. Chromatogr. B 877 (2009) 2224–2234.
[15] (a) EurachemGuide: The Fitness for Purpose of Analytical Methods,
Eurachem, Teddington, 1998. (b) B. Magnusson and U. Örnemark (Eds.)
Eurachem Guide: The Fitness for Purpose of Analytical Methods – A
Laboratory Guide to Method Validation and Related Topics, (second ed. 2014).
ISBN 978-91-87461-59-0. Available from www.eurachem.org.
[16] Harmonized Guidelines for single-laboratory validation of method of
analyses (IUPAC Technical Report), Pure Appl. Chem. 74(5) (2002) 835–855.
[17] ICH harmonized tripartite guideline: validation of analytical procedures: text
and methodology Q2(R1), International Conference of harmonization of
technical requirements for registration of pharmaceuticals for human use,
2005.
[18] (a)Guidance for Industry: Bioanalytical method validation, U.S. Department
of Health and Human Services Food and Drug Administration, 2001.
(b) Guidance for Industry: Bioanalytical method validation, Draft Guidance, U.
S. Department of Health and Human Services Food and Drug
Administration
http://www.fda.gov/downloads/Drugs/
GuidanceComplianceRegulatoryInformation/Guidances/UCM368107.pdf,
2013.
[19] Guidance on Bioanalytical Method Validation, European Medicines Agency,
2011.
[20] AOAC Guidelines for Single-Laboratory Validation of Chemical Methods for
Dietary Supplements and Botanicals, Official Methods of Analysis, nineteenth
ed., Appendix K, AOAC, INTERNATIONAL, Gaithersburg, MD, 2012.
[21] S. Chandran, R.S.P. Singh, Comparison of various international guidelines for
analytical method validation, Pharmazie 62 (2007) 4–14.
[22] R. Bonfilio, E.C.L. Cazedey, M.B. de Araújo, H.R.N. Salgado, Analytical
validation of quantitative high-performance liquid chromatographic
methods in pharmaceutical analysis: a practical approach, Crit. Rev. Anal.
Chem. 42 (2012) 87–100.
[23] E. Rozet, R.D. Marini, E. Ziemons, B. Boulanger, P. Hubert, Advances in
validation, risk and uncertainty assessment of bioanalytical methods, J.
Pharm. Biomed. 55 (2011) 848–858.
[24] G.A. Shabir, Validation of high-performance liquid chromatography methods
for pharmaceutical analysis: understanding the differences and similarities
between validation requirements of the US Food and Drug Administration,
the US Pharmacopeia and the International Conference on Harmonization, J.
Chromatogr. A 987 (2003) 57–66.
[25] F.T. Peters, Method validation using LC–MS, in: A. Polettini (Ed.), Applications
Liquid
Chromatography–Mass
Spectrometry
in
Toxicology,
of
Pharmaceutical Press, London, 2006, pp. 71–96.
[26] NordVal Protocol No. 2, Guide in validation of alternative propietary chemical
methods, 2010.
[27] O. González, M.E. Blanco, G. Iriarte, L. Bartolomé, M.I. Maguregui, R.M. Alonso,
Bioanalytical chromatographic method validation according to current
regulations with a special focus on the non-well defined parameters limit
of quantification, robustness and matrix effect, J. Chromatogr. A 1353 (2014)
10–27.
[28] International Vocabulary of Metrology–Basic and General Concepts and
Associated Terms (VIM), JCGM 200:2012.
[29] A. Kruve, R. Rebane, K. Kipper, M.-L. Oldekop, H. Evard, K. Herodes, P. Ravio, I.
Leito, Tutorial review on validation of liquid chromatography–mass
spectrometry methods: Part II Anal. Chim. Acta, http://dx.doi.org/10.1016/
j.aca.2015.02.016.
[30] R.D. Smith, J.A. Loo, C.G. Endmonds, C.J. Barinaga, H.R. Udseth, New
developments in biochemical mass spectrometry: electrospray ionization,
Anal. Chem. 62 (9) (1990) 882–899.
[31] N.B. Cech, C.G. Enke, Practical implications of some recent studies in
electrospray ionization fundamentals, Mass Spectrom. Rev. 20 (2001) 362–387.
[32] R. Kostiainen, T. Kotiaho, T. Kuuranne, S. Auriola, Liquid chromatography/
atmospheric pressure ionization-mass spectrometry in drug metabolism
studies, J. Mass Spectrom. 38 (4) (2003) 357–372.
[33] A. Raffaell, A. Saba, Atmospheric pressure photoionization mass
spectrometry, Mass Spectrom. Rev. 22 (5) (2003) 318–331.
[34] D.B. Robb, M.W. Blades, State-of-the-art in atmospheric pressure
photoionization for LC/MS, Anal. Chem. 627 (2008) 34–49.
[35] A. Kruve, K. Kaupmees, J. Liigand, I. Leito, Sodium adduct formation efficiency
in ESI source, J. Mass Spectrom. 48 (2013) 695–702.
[36] B.K. Choi, A.I. Gusev, D.M. Hercules, Postcolumn introduction of an internal
standard for quantitative LC–MS analysis, Anal. Chem. 71 (1999)
4107–4110.
[37] F.M. Lagerwerf, W.D. van Dongen, R.J.J.M. Steenvoorden, M. Honing, J.H.G.
Jonkman, Exploring the boundaries of bioanalytical quantitative LC–MS–MS,
Trends Anal. Chem. 19 (7) (2000) 418–427.
[38] W.M.A. Niessen, P. Manini, R. Andreoli, Matrix effects in quantitative
pesticide analysis using liquid chromatography–mass spectrometry, Mass
Spectrom. Rev. 25 (2006) 881–899.
[39] G.G. Guilbault, M. Hjelm, Nomenclature for automated and mechanised
analysis, Pure Appl. Chem. 61 (1989) 1657–1664.
[40] B.K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng, Strategies for the
assessment of matrix effect in quantitative bioanalytical methods based on
HPLC-MS/MS, Anal. Chem. 75 (2003) 3019–3030.
[41] P.J. Taylor, Matrix effects: the Achilles heel of quantitative high-performance
liquid chromatography–electrospray–tandem mass spectrometry, Clin.
Biochem. 38 (2005) 328–334.
[42] H. Trufelli, P. Palma, G. Famiglini, A. Capiello, An overview of matrix effects in
liquid chromatography–mass spectrometry, Mass Spectrom. Rev. 30 (2011)
491–509.
[43] H. Mei, Y. Hsieh, C. Nardo, X. Xu, S. Wang, K. Ng, W.A. Korfmacher,
Investigation of matrix effects in bioanalytical high-performance liquid
chromatography/tandem mass spectrometric assays: application to drug
discovery, Rapid Commun. Mass Spectrom. 17 (2003) 97–103.
[44] L. Alder, S. Lüderitz, K. Lindtner, S.-H. Stan, The ECHO technique–the more
effective way of data evaluation in liquid chromatography–tandem mass
spectrometry analysis, J. Chromatogr. A 1058 (2004) 67–79.
[45] A. Kruve, K. Herodes, I. Leito, Combating matrix effects in LC/ESI/MS: the
extrapolative dilution approach, Anal. Chim. Acta 651 (2009) 75–80.
[46] A. Kruve, I. Leito, Comparison of different methods aiming to account for/
overcome matrix effects in LC/ESI/MS on the example of pesticide analyses,
Anal. Methods 5 (2013) 3035–3044.
[47] R. Bonfiglio, R.C. King, T.V. Olah, K. Merkle, The effects of sample preparation
methods on the variability of the electrospray ionization response for model
drug compounds, Rapid Commun. Mass Spectrom. 13 (1999) 1175–1185.
[48] S. Souverain, S. Rudaz, J.-L. Veuthey, Matrix effect in LC–ESI–MS and LC–
APCI–MS with off-line and on-line extraction procedure, J. Chromatogr. A
1058 (2004) 61–66.
[49] R. Dams, M.A. Huestis, W.E. Lambert, C.M. Murphy, Matrix effect in bioanalysis of illicit drugs with LC-MS/MS: influence of ionization type, sample
preparation, and biofluid, J. Am. Soc. Mass Spectrom. 14 (2003) 1290–1294.
[50] P. Keski-Rahkonen, K. Huhtinen, R. Desai, D.T. Harwood, D.J. Handelsman, M.
Poutanen, S. Auriola, LC–MS analysis of estradiol in human serum and
endometrial tissue: comparison of electrospray ionization, atmospheric
pressure chemical ionization and atmospheric pressure photoionization, J.
Mass Spectrom. 48 (2013) 1050–1058.
[51] O.A. Ismaiel, M.S. Halquist, M.Y. Elmamly, A. Shalaby, H.T. Karnes, Monitoring
phospholipids for assessment of ion enhancement and ion suppression in ESI
and APCI LC/MS/MS for chlorpheniramine in human plasma and the
importance of multiple source matrix effect evaluations, J. Chromatogr. B
875 (2008) 333–343.
[52] E. Beltrán, M. Ibánez, J.V. Sancho, F. Hernández, Determination of patulin in
apple and derived products by UHPLC-MS/MS. Study of matrix effects with
atmospheric pressure ionisation sources, Food Chem. 142 (2014) 400–407.
[53] E.M. Thurman, I. Ferrer, D. Barcelo, Choosing between atmospheric pressure
chemical ionization and electrospray ionization interfaces for the HPLC/MS
analysis of pesticides, Anal. Chem. 73 (2001) 5441–5449.
[54] A. Kloepfer, J.B. Quintana, T. Reemtsma, Operational options to reduce matrix
in
liquid
chromatography–electrospray
ionization-mass
effects
spectrometry analysis of aqueous environmental samples, J. Chromatogr.
A 1067 (2005) 153–160.
[55] E. Chambers, D.M. Wagrowski-Diehl, Z. Lu, J.R. Mazzeo, Systematic and
comprehensive strategy for reducing matrix effects in LC/MS/MS analyses, J.
Chromatogr. B 852 (2007) 22–34.
[56] J.V. Sancho, O.J. Pozo, F.J. Lopez, F. Hernandez, Different quantitation
approaches for xenobiotics in human urine samples by liquid
chromatography/electrospray tandem mass spectrometry, Rapid Commun.
Mass Spectrom. 16 (2002) 639–645.
[57] A.I. Garcia-Valcarcel, J.L. Tadeo, A combination of ultrasonic assisted
extraction with LC-MS/MS for the determination of organophosphorus
pesticides in sludge, Anal. Chim. Acta 641 (2009) 117–123.
[58] H. Stahnke, S. Kittlaus, G. Kempe, L. Alder, Reduction of matrix effects in liquid
chromatography-electrospray ionization-mass spectrometry by dilution of
the sample extracts: how much dilution is needed? Anal. Chem. 84 (2012)
1474–1484.
[59] J. Zrostlikova, J. Hajšlova, J. Pouska, P. Begany, Alternative calibration
approaches to compensate the effect of co-extracted matrix components in
liquid chromatography–electrospray ionisation tandem mass spectrometry
analysis of pesticide residues in plant materials, J. Chromatogr. A 973 (2002)
13–26.
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
[60] A. Kaufmann, P. Butcher, Segmented post-column analyte addition; a concept
for continuous response control of liquid chromatography/mass
spectrometry peaks affected by signal suppression/enhancement, Rapid
Commun. Mass Spectrom. 19 (5) (2005) 611–617.
[61] K.K. Murray, R.K. Boyd, M.N. Eberlin, G.J. Langley, L. Li, Y. Naito, Definitions of
terms relating to mass spectrometry (IUPAC recommendations 2013), Pure
Appl. Chem. 85 (7) (2013) 1515–1609.
[62] A. Singleton, Recent advances in bioanalytical sample preparation for LC–MS
analyses, Bioanalysis 4 (9) (2012) 1123–1140.
[63] A. Tan, S. Hussain, A. Musuku, R. Masse, Internal standard response variations
during incurred sample analysis by LC-MS/MS: case by case troubleshooting, J. Chromatogr. B 877 (27) (2009) 3201–3209.
[64] F.T. Peters, O.H. Drummer, F. Musshoff, Validation of new methods, Forensic
Sci. Int. 165 (2007) 216–224.
[65] SANCO/12571/2013, Guidance document on analytical quality control and
validation procedures for pesticide residues analysis in food and feed. 2013.
[66] European Commission Decision 2002/657/EC implementing Council
Directive 96/23/EC concerning the performance of analytical methods and
the interpretation of results, Off. J. Eur. Commun. L221 (2002) 8–36.
[67] E.I. Miller, F.M. Wylie, J.S. Oliver, Simultaneous detection and quantification
of amphetamines, diazepam and its metabolites cocaine and its metabolites,
and opiates in hair by LC–ESI-MS–MS using a single extraction method, J.
Anal. Toxicol. 32 (2008) 457–469.
[68] A.V. Ramani, P. Sengupta, R. Mullangi, Development and validation of a highly
sensitive and robust LC–ESI–MS/MS method for simultaneous quantitation of
simvastatin acid, amlodipine and valsartan in human plasma: application to a
clinical pharmacokinetic study, Biomed. Chromatogr. 23 (2009) 615–622.
[69] J. Feng, X.-W. Yang, R.-B. Huang, H.-Y. Zhang, M. He, Q.-C. Huang,
Development and validation of an LC–ESI–MS/MS method for the
determination of nitidine chloride in rat plasma, J. Chromatogr. B 887-888
(2012) 43–47.
[70] J. Beyer, F.T. Peters, T. Kraemer, H. Maurer, Detection and validated
quantification of nine herbal phenalkylamines and methcathinone in human
blood plasma by LC-MS/MS with electrospray ionization, J. Mass Spectrom.
42 (2007) 150–160.
[71] A. Gupta, P. Singhal, P.S. Shrivastav, M. Sanyal, Application of a validated ultra
performance liquid chromatography–tandem mass spectrometry method for
the quantification of darunavir in human plasma for a bioequivalence study
in Indian subjects, J. Chromatogr. B 879 (2011) 2443–2453.
[72] H.N. Mistri, A.G. Jangid, A. Pudage, N. Gomes, M. Sanyal, P. Shrivastav, High
throughput LC-MS/MS method for simultaneous quantification of
lamivudine, stavudine and nevirapine in human plasma, J. Chromatogr. B
879 (2007) 320–332.
[73] H.F. Schröder, W. Gebhardt, M. Thevis, Anabolic doping, and lifestyle drugs,
and selected metabolites in wastewater-detection, quantification, and
behaviour monitored by high-resolution MS and MSn before and after
sewage treatment, Anal. Bioanal. Chem. 398 (2010) 1207–1229.
[74] D. Yeniceli, E. Sener, O.T. Korkmaz, D. Dogrukol-Ak, N. Tuncel, A simple and
sensitive LC–ESI–MS (ion trap) method for the determination of bupropion
and its major metabolite, hydroxybupropion in rat plasma and brain
microdialysates, Talanta 84 (2011) 19–26.
[75] F.P. Gomes, P.L. García, J.M.P. Alves, A.K. Singh, E.R.M. Kedor-Hackmann, M.I.R.
M. Santoro, Development and validation of stability-indicating HPLC
methods for quantitative determination of pravastatin fluvastatin,
atorvastatin, and rosuvastatin in pharmaceuticals, Anal. Lett. 42 (2009)
1784–1804.
[76] F. Calbiani, M. Careri, L. Elviri, A. Mangia, L. Pistara, I. Zagnoni, Development
and in-house validation of a liquid chromatography–electrospray–tandem
mass spectrometry method for the simultaneous determination of Sudan I,
Sudan II Sudan III and Sudan IV in hot chilli products, J. Chromatogr. A 1042
(2004) 123–130.
[77] A. Angioni, L. Porcu, F. Pirisi, LC/DAD/ESI/MS Method for the determination of
imidacloprid thiacloprid, and spinosad in olives and olive oil after field
treatment, J. Agric. Food Chem. 59 (2011) 11359–11366.
[78] J. Ascari, S. Dracz, F.A. Santos, J.A. Lima, M.H.G. Diniz, E.A. Vargas, Validation of
an LC-MS/MS method for malachite green (MG) leucomalachite green (LMG),
crystal violet (CV) and leucocrystal violet (LCV) residues in fish and shrimp,
Food Addit. Contam. 29 (2012) 602–608.
[79] M. Careri, C. Corradini, L. Elviri, I. Nicoletti, I. Zagnoni, Liquid
chromatography-electrospray tandem mass spectrometry of cis-resveratrol
and trans-resveratrol: development, validation, and application of the
method to red wine, grape, and winemaking byproducts, J. Agric. Food Chem.
52 (2004) 6868–6874.
[80] H.G.F. Smienk, D. Calvo, P. Razquin, E. Domínguez, L. Mata, Single laboratory
validation of a ready-to-use phosphatase inhibition assay for detection of
okadaic acid toxins, Toxins 4 (2012) 339–351.
[81] B. Cordero, F. Canale, D. Del Rio, C. Bicchi, Identification quantitation,and
method validation for flavan-3-ols in fermented ready-to-drink teas from the
Italian market using HPLC-UV/DAD and LC-MS/MS, J. Sep. Sci. 32 (2009)
3643–3651.
[82] A.J. Giron, K. Deventer, P. Van Eenoo, Development and validation of an open
screening method for diuretics, stimulants and selected compounds in
human urine by UHPLC–HRMS for doping control, Anal. Chim. Acta 721
(2012) 137–146.
[83] M. Mattarozzi, M. Milioli, C. Cavalieri, F. Bianchi, M. Careri, Rapid desorption
electrospray ionization-high resolution mass spectrometry method for the
[84]
[85]
[86]
[87]
[88]
[89]
[90]
[91]
[92]
[93]
[94]
[95]
[96]
[97]
[98]
[99]
[100]
[101]
[102]
[103]
[104]
[105]
[106]
[107]
43
analysis of melamine migration from melamine tableware, Talanta 10 (2012)
453–459.
P. Montoro, M. Maldini, M. Russo, S. Postorino, S. Piacente, C. Pizza, Metabolic
profiling of roots of liquorice (Glycyrrhiza glabra) from different geographical
areas by ESI/MS/MS and determination of major metabolites by LC–ESI/MS
and LC–ESI/MS/MS, J. Pharm. Biomed. Anal. 54 (2011) 535–544.
M.J. Rumpler, Quantitative analysis of 11-nor-9-carboxytetrahydrocannbinol(THC-COOH) in urine by LC-MS/MS following a simple
filtration, J. Chromatogr. B 957 (2014) 77–83.
N. Venisse, C. Grignon, B. Brunet, S. Thévenot, A. Bacle, V. Migeot, A. Dupuis,
Reliable quantification of bisphenol A and its chlorinated derivatives in
human urine using UPLC-MS/MS method, Talanta 125 (2014) 284–292.
E.N. Ediage, J.D. Di Mavungu, S. Monbaliu, C. Van Peteghem, S. De Saeger, A
validated multianalyte LC-MS/MS method for quantification of
25 mycotoxins in cassava flour, peanut cake and maize samples, J. Agric.
Food Chem. 9 (2011) 5173–5180.
S.C. Utture, K. Banerjee, S. Dasgupta, S.H. Patil, M.R. Jadhav, S.S. Wagh, S.S.
Kolekar, M.A. Anuse, P.G. Adsule, Dissipation and Distribution Behavior of
Azoxystrobin, Carbendazim, and Difenoconazole in Pomegranate Fruits, J.
Agric. Food Chem. 59 (2011) 7866–7873.
M.M. De Zan, M.S. Cámara, J.C. Robles, S.V. Kergaravat, H.C. Goicoechea,
Development and validation of a simple stability-indicating high performance
liquid chromatographic method for the determination of miconazole nitrate in
bulk and cream formulations, Talanta 79 (2009) 762–767.
C. Kmellár, P. Fodor, L. Pareja, C. Ferrer, M.A. Martínez-Uroz, A. Valverde, A.R.
Fernandez-Alba, Validation and uncertainty study of a comprehensive list of
160 pesticide residues in multi-class vegetables by liquid chromatography–
tandem mass spectrometry, J. Chromatogr. A 1215 (2008) 37–50.
K.P. Yánez, M.T. Martin, J.L. Bernal, M.J. Nozal, J. Bernal, Determination of
spinosad at trace levels in bee pollen and beeswax with solid–liquid
extraction and LC–ESI-MS, J. Sep. Sci. 37 (2014) 204–210.
R.N. Rao, P.K. Maurya, M. Ramesh, R. Srinivas, S.B. Agwane, Development of a
validated high-throughput LC–ESI–MS method for determination of
sirolimus on dried blood spots, Biomed. Chromatogr. 24 (2010) 1356–1364.
M.J. Bogusz, H. Hassan, E. Al-Enazi, Z. Ibrahim, M. Al-Tufail, Application of LC–
ESI–MS–MS for detection of synthetic adulterants in herbal remedies, J.
Pharm. Biomed. Anal. 41 (2006) 554–564.
H. Yamaguchi, A. Fujikawa, H. Ito, T. Tanaka, A. Furugen, K. Miyamori, N.
Takahashi, J. Ogura, M. Kobayashi, T. Yamada, N. Mano, K. Iseki, A rapid and
sensitive LC/ESI–MS/MS method for quantitative analysis of docetaxel in
human plasma and its application to a pharmacokinetic study, J. Chromatogr.
B 893–894 (2012) 157–161.
A. Claus, G.M. Weisz, D.R. Kammerer, R. Carle, A. Schieber, A method for the
determination of acrylamide in bakery products using ion trap LC–ESI–MS/
MS, Mol. Nutr. Food Res. 49 (2005) 918–925.
M. Gros, M. Petrovi
c, D. Barceló, Development of a multi-residue analytical
methodology based on liquid chromatography–tandem mass spectrometry
(LC-MS/MS) for screening and trace level determination of pharmaceuticals
in surface and wastewaters, Talanta 70 (2006) 678–690.
A. Sannino, L. Bolzoni, M. Bandini, Application of liquid chromatography with
electrospray tandem mass spectrometry to the determination of a new
generation of pesticides in processed fruits and vegetables, J. Chromatogr. A
1036 (2004) 161–169.
J. Wang, Determination of Five Macrolide Antibiotic Residues in Honey by LC–
ESI–MS and LC–ESI–MS/MS, J. Agric. Food Chem. 52 (2004) 171–181.
G.A. Shabir, W.J. Lough, S.A. Arain, T.K. Bradshaw, Evaluation and application
of best practice in analytical method validation, J. Liq. Chromatogr. Relat.
Technol. 20 (2007) 311–333.
E. Rozet, A. Ceccato, C. Hubert, E. Ziemons, R. Oprean, S. Rudaz, B. Boulanger, P.
Hubert, Analysis of recent pharmaceutical regulatory documents on
analytical method validation, J. Chromatogr A. 1158 (2007) 111–125.
IUPAC, Compendium of Chemical Terminology, 2nd ed., (the Gold Book).
Compiled by A.D. McNaught, A. Wilkinson. Blackwell Scientific Publications,
Oxford (1997). XML on-line corrected version: http://goldbook.iupac.org
(2006-) created by M. Nic, J. Jirat, B. Kosata, updates compiled by A. Jenkins.
ISBN 0-9678550-9-8. doi:10.1351/goldbook, last update: 24-02-2014;
version: 2.3.3.
C.J. Anagnostopoulos, P. Aplada Sarli, K. Liapis, S.A. Haroutounian, G.E.
Miliadis, Validation of two variations of the QuEChERS method for the
determination of multiclass pesticide residues in cereal-based infant foods
by LC-MS/MS, Food Anal. Method 5 (2012) 664–683.
S.D. Richardson, Environmental mass spectrometry: emerging contaminants
and current issues, Anal. Chem. 82 (12) (2010) 4742–4774.
R. Diaz, M. Ibáñez, J.V. Sancho, F. Hernández, Qualitative validation of a liquid
chromatography–quadrupole-time of flight mass spectrometry screening
method for organic pollutants in waters, J. Chromatogr. A 1276 (2013) 47–57.
J. Nácher-Mestre, M. Ibáñez, R. Serrano, J. Pérez-Sánchez, F. Hernández,
Qualitative screening of undesirable compounds from feeds to fish by liquid
chromatography coupled to mass spectrometry, J. Agric. Food Chem. 61 (9)
(2013) 2077–2087.
R. Macarthur, C. von Holst, A protocol for the validation of qualitative
methods of detection, Anal. Methods 4 (2012) 2744–2754.
H.G. Mol, P. Zomer, M. de Koning, Qualitative aspects and validation of a
screening method for pesticides in vegetables and fruits based on liquid
chromatography coupled to full scan high resolution (orbitrap) mass
spectrometry, Anal. Bioanal. Chem. 403 (10) (2012) 2891–2908.
44
A. Kruve et al. / Analytica Chimica Acta 870 (2015) 29–44
[108] H. Kadar, B. Veyrand, J.P. Antignac, S. Durand, F. Monteau, B. Le Bizac,
Comparative study of low- versus high-resolution liquid chromatographymass spectrometric strategies for measuring perfluorinated contaminants in
fish, Food Addit. Contam. A 28 (9) (2011) 1261–1273.
[109] R.J. Singh, R.L. Taylor, G. Satyanarayana Reddy, S.K.G. Grebe, C-3 Epimers can
account for a Significant proportion of total circulating 25-hydroxyvitamin D
in infants, complicating accurate measurement and interpretation of vitamin
D status, J. Clin. Endocr. Metab. 91 (8) (2006) 3055–3061.
[110] A. Gentili, F. Caretti, S. Bellante, L.M. Rocca, R. Curini, A. Venditti,
Development and validation of two multiresidue liquid chromatography
tandem mass spectrometry methods based on a versatile extraction
procedure for isolating non-steroidal anti-inflammatory drugs from
bovine milk and muscle tissue, Anal. Bioanal. Chem. 404 (2012) 1375–1388.
[111] A. Domenech, N. Cortes-Francisco, O. Palacios, J.M. Franco, P. Riobo, J.J.
Lierena, S. Vichi, J. Caixach, Determination of lipophilic marine toxins in
mussels. Quantification and confirmation criteria using high resolution mass
spectrometry, J. Chromatogr. A 1328 (2014) 16–25.
[112] Ó.J. Pozo, J.V. Sancho, M. Ibáñez, F. Hernández, W.M.A. Niessen, Confirmation
of organic micropollutants detected in environmental samples by liquid
chromatography tandem mass spectrometry: achievements and pitfalls,
TrAC Trends Anal Chem. 25 (10) (2006) 1030–1042.
[113] R. Bethem, J. Boison, J. Gala, D. Heller, S. Lehotay, J. Loo, S. Musser, P. Price, S.
Stein, Establishing the fitness for purpose of mass spectrometric methods, J.
Am. Soc. Mass Spectrom. 14 (2003) 528–541.
[114] J.A. Van Leeuwen, L.M.C. Buydens, B.G.M. Vandeginste, G. Kateman, P.J.
Schoenmakers, M. Mulholland, RES, an expert system for the set-up and
interpretation of a ruggedness test in HPLC method validation. Part 1: the
ruggedness test in HPLC method validation, Chemom. Intell. Lab. Syst. 10
(1991) 337–347.
[115] E. Karageorgou, V. Samanidou, Youden test application in robustness assays
during method validation, J. Chromatogr. A 65 (2014) 234–240.
[116] W.J. Youden, E.H. Steiner, Statistical manual of the association of official, Anal.
Chem. (1975) 50–55.
[117] R. Boque, A. Maroto, J. Riu, X. Rius, Validation of analytical methods, Grasas
Aceites 53 (1) (2002) 128–143.
[118] V.V. Heyden, A. Nijhuis, J. Smeyers-Verbeke, B.G.M. Vandeginste, D.L.
Massart, Guidance for robustness/ruggedness tests in method validation, J.
Pharm. Biomed. Anal. 24 (2001) 723–753.
[119] L. Cuadros-Rodriguez, R. Romero, J.M. Bosque-Sendra, Studies in research and
development of analytical processes, Crit. Rev. Anal. Chem. 35 (1) (2005)
57–69.
[120] J. Caporal-Gautier, J.M. Nivet, P. Algranti, M. Guilloteau, M. Histe, M. Lallier, J.J.
N’Guyen-Huu, R. Russotto, Guide de validation analytique, rapport d’une
commission SFSTP, S. T. P. Pharma. Pratiques 2 (1992) 205–239.
[121] The United States Pharmacopeia, 23rd ed., National Formulary 18, United
States Pharmacopeial Convention, Rockville, USA, 1995.
[122] The United States Pharmacopeia, 35th ed., Rockville, The United States
Pharmacopeial Convention, Inc., 2012, 3261–3279.
[123] E. Aguilera, R. Lucena, S. Cardenas, M. Valcarcel, E. Trullols, I. Ruisanchez,
Robustness in qualitative analyses: a practical approach, Trac-Trend Anal.
Chem. 25 (6) (2006) 621–627.
[124] B. Sundström, G. Johansson, H. Kokkonen, T. Cederholm, S. Wållberg-Jonsson,
Plasma phospholipid fatty acid content is related to disease activity in
ankylosing spondylitis, J. Rheumatol. 39 (2012) 327–333.
[125] A.C. Thiébaut, M. Rotival, E. Gauthier, G.M. Lenoir, M.C. Boutron-Ruault, V.
Joulin, F. Clavel-Chapelon, V. Chajès, Correlation between serum
phospholipid fatty acids and dietary intakes assessed a few years earlier,
Nutr. Cancer 61 (2009) 500–509.
[126] C.E. Childs, M. Romeu-Nadal, G.C. Burdge, P.C. Calder, The polyunsaturated
fatty acid composition of hepatic and plasma lipids differ by both sex and
dietary fat intake in rats, J. Nutr. 140 (2010) 245–250.
[127] P. Bastos-Amador, F. Royo, E. Gonzalez, J. Conde-Vancells, L. Palomo-Diez, F.E.
Borras, J.M. Falcon-Perez, Proteomic analysis of microvesicles from plasma of
healthy donors reveals high individual variability, J. Proteomics 75 (2012)
3574–3584.
[128] M.E. Dasenaki, N.S. Thomaidis, Multi-residue determination of seventeen
sulfonamides and five tetracyclines in fish tissue using a multi-stage LC–ESI–
MS/MS approach based on advanced mass spectrometric techniques, Anal.
Chim. Acta 672 (2010) 93–102.
[129] B. Dejaegher, X. Capron, J. Smeyers-Verbeke, Y. Vander Heyden,
Randomization tests to identify significant effects in experimental design for
robustness testing, Anal. Chim. Acta 564 (2006) 184–200.
[130] R.L. Plackett, J.P. Burman, The design of optimum multifactorial experiments,
Biometrica 33 (4) (1946) 305–325.
[131] Y. Vander Heyden, K. Luypaert, C. Hartmann, D.L. Massart, J. Hoogmartens, J.
De Beer, Ruggedness tests on the high-performance liquid chromatography
assay of the United States Pharmacopeia XXII for tetracycline hydrochloride.
A comparison of experimental designs and statistical interpretations, Anal.
Chim. Acta 312 (1995) 245–262.
[132] Y. Vander Heyden, C. Hartmann, D.L. Massart, A.M.J. Hollands, P. Nuyten, P.
Schoenmakers, Ruggedness testing of a size-exclusion chromatographic
assay for low-molecular-mass polymers, J. Chromatogr. A 756 (1) (1996)
89–106.
[133] B. Dejaegher, Y. Vander Heyden, Ruggedness and robustness testing, J.
Chromatogr. A 1158 (2007) 138–157.
[134] C. Benetti, R. Piro, G. Binato, R. Angeletti, G. Biancotto, Simultaneous
determination of lincomycin and five macrolide antibiotic residues in honey
by liquid chromatography coupled to electrospray ionization mass
spectrometry (LC-MS/MS), Food Addit. Contam. 23 (11) (2006) 1099–1108.
[135] P. Jedziniak, T. Szprengier-Juszkiewicz, M. Olejnik, Determination of
benzimidazoles and levamisole residues in milk by liquid chromatography–
mass spectrometry: screening method development and validation, J.
Chromatogr. A 1216 (2009) 8165–8172.
[136] R. Zeleny, S. Harbeck, H. Schimmel, Validation of a liquid chromatography–
tandem mass spectrometry method for the identification and quantification
of 5-nitroimidazole drugs and their corresponding hydroxy metabolites in
lyophilised pork meat, J. Chromatogr. A 1216 (2009) 249–256.
[137] M. Gasparini, M. Curatolo, W. Assini, E. Bozzoni, N. Tognoli, G. Dusi,
Confirmatory method for the determination of nandrolone and trenbolone in
urine samples using immunoaffinity cleanup and liquid chromatography–
tandem mass spectrometry, J. Chromatogr. A 1216 (2009) 8059–8066.
[138] C.C.G.O. Lopes, H.R.N. Salgado, Development of a validated stabilityindicating LC assay and stress degradation studies of linezolid in tablets,
Chromatographia 69 (2009) S129–S135.
[139] N. Kaul, H. Agrawal, A.R. Paradkar, K.R. Mahadik, The ICH guidance in
practice: stress degradation studies on indinavir sulphate and development
of a validated specific stability-indicating HPTLC assay method, Il Farmaco 59
(2004) 729–738.
[140] L.B. Junior, F.T. Uchoa, S.S. Guterres, T.D. Costa, Development and validation of
LC-MS/MS method for the simultaneous determination of quinine and
doxycycline in pharmaceutical formulations, J. Liq. Chromatogr. Relat.
Technol. 1 (2009) 2699–2711.
[141] C. Yardimci, N. Özaltın, Method development and validation for the
simultaneous determirnation of rosiglitazone and metformin in
pharmaceutical preparations by capillary zone electrophoresis, Anal.
Chim. Acta 549 (2005) 88–95.
[142] S. Thomas, S. Shandilya, A. Bharati, S.K. Paul, A. Agarwal, C.S. Mathela,
Identification, characterization and quantification of new impurities by
LC–ESI/MS/MS and LC–UV methods in rivastigmine tartrate active
pharmaceutical ingredient, J. Pharm. Biomed. Anal. 57 (2012) 39–51.
[143] C. von Holst, A. Müller, E. Björklund, E. Anklam, In-house validation of a
simplified method for the determination of PCBs in food and feedingstuffs,
Eur. Food Res. Technol. 213 (2001) 154–160.
[144] N.A. Épshtein, Validation of HPLC techniques for pharmaceutical analysis,
Pharm. Chem. J. 38 (2004) 212–228.
[145] T. Mirza, H.S.I. Tan, Determination of captopril in pharmaceutical tablets by
anion-exchange HPLC using indirect photometric detection; a study in
systematic method development, J. Pharm. Biomed. Anal. 25 (1) (2001) 39–52.
[146] G. Szekely, B. Henriques, M. Gil, C. Alvarez, Experimental design for the
optimization and robustness testing of a liquid chromatography tandem
mass spectrometry method for the trace analysis of the potentially genotoxic
1,3-diisopropylurea, Drug Test. Anal. 6 (2014) 898–908.
[147] F. Calbiani, M. Careri, L. Elviri, A. Mangia, I. Zagnoni, Validation of an ion-pair
liquid chromatography-electrospray-tandem mass spectrometry method for
the determination of heterocyclic aromatic amines in meat-based infant
foods, Food Addit. Contam. 24 (8) (2007) 833–841.
[148] L.A. Currie, Limits for qualitative detection and quantitative determination,
Anal. Chem. 40 (1968) 586–593.
[149] H. Evard, A. Kruve, I. Leito, LoD determination in LC/MS/MS: choosing the
appropriate methodology and comparability of the results, J. Chromatogr. A
(2015) submitted for publication.
[150] J. Fonollosa, A. Vergara, R. Huerta, S. Marco, Estimation of the limit of detection
using information theory measures, Anal. Chim. Acta 810 (2014) 1–9.
[151] W. Huber, Basic calculations about the limit of detection and its optimal
determination, Accred. Qual. Assur. 8 (2003) 213–217.
[152] N. Lindegardh, W. Hanpithakpong, Y. Wattanagoon, P. Singhasivanon, N.J.
White, N.P.J. Day, Development and validation of a liquid chromatographic–
tandem mass spectrometric method for determination of oseltamivir and its
metabolite oseltamivir carboxylate in plasma, saliva and urine, J. Chromatogr.
B 859 (2007) 74–78.
[153] L.A. Currie, Nomenclature in evaluation of analytical methods including
detection and quantification capabilities, Pure Appl. Chem. 67 (1995)
1699–1723.
[154] Directive 2000/60/EC of the European Parliament and of the Council
establishing a framework for Community action in the field of water policy,
CELEX-EUR. Off. J. L 327 (2000) 1–72.
[155] ISO 11843–6:2013 Capability of detection – Part 6: Methodology for the
determination of the critical value and the minimum detectable value in
Poisson distributed measurements by normal approximations.
[156] S. Szilagyi, B. de la Calle, Development and validation of an analytical method
for the determination of semicarbazide in fresh egg and in egg powder based
on the use of liquid chromatography tandem mass spectrometry, Anal. Chim.
Acta 572 (2006) 113–120.
[157] M. Rokka, K. Peltonen, Simultaneous determination of four coccidiostats in
eggs and broiler meat: validation of an LC-MS/MS method, Food Addit.
Contam. 23 (2006) 470–478.