You are on page 1of 15

Toxicology Mechanisms and Methods

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/itxm20

Dispersive liquid-liquid microextraction (DLLME)


for determination of tricyclic antidepressants in
whole blood and plasma samples and analysis by
liquid chromatography with diode array detector
(LC-DAD)

Dener Gomes Berlato, André Lucas Bezerra Pacheco, Gustavo Andrade


Ugalde, Fernanda Ziegler Reginato, Geovane de Almeida Saldanha, Tiago
Franco de Oliveira, Sarah Eller & André Valle de Bairros

To cite this article: Dener Gomes Berlato, André Lucas Bezerra Pacheco, Gustavo Andrade
Ugalde, Fernanda Ziegler Reginato, Geovane de Almeida Saldanha, Tiago Franco de Oliveira,
Sarah Eller & André Valle de Bairros (13 Oct 2023): Dispersive liquid-liquid microextraction
(DLLME) for determination of tricyclic antidepressants in whole blood and plasma samples and
analysis by liquid chromatography with diode array detector (LC-DAD), Toxicology Mechanisms
and Methods, DOI: 10.1080/15376516.2023.2269236

To link to this article: https://doi.org/10.1080/15376516.2023.2269236

Published online: 13 Oct 2023.

Submit your article to this journal

Article views: 12

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=itxm20
TOXICOLOGY MECHANISMS AND METHODS
https://doi.org/10.1080/15376516.2023.2269236

RESEARCH ARTICLE

Dispersive liquid-liquid microextraction (DLLME) for determination of tricyclic


antidepressants in whole blood and plasma samples and analysis by liquid
chromatography with diode array detector (LC-DAD)
Dener Gomes Berlatoa, Andre
� Lucas Bezerra Pachecoa, Gustavo Andrade Ugaldea, Fernanda Ziegler Reginatoa,
Geovane de Almeida Saldanhaa, Tiago Franco de Oliveirab, Sarah Ellerb and Andr�e Valle de Bairrosa
a
Nucleous Applied to Toxicology (NAT), Department of Clinical and Toxicological Analysis, Federal University of Santa Maria, Santa Maria,
Brazil; bGraduate Program in Health Sciences, Federal University of Health Sciences of Porto Alegre, RS, Brazil

ABSTRACT ARTICLE HISTORY


Microextractions have been developed for the tricyclic antidepressants (TCAs) analysis in biological matri­ Received 26 July 2023
ces, including dispersive liquid-liquid microextraction (DLLME). The proposed DLLME employed 490 mL of Revised 26 September 2023
biological sample (whole blood or plasma), which were added 15 mg of NaCl, 10 mL of medazepam as Accepted 5 October 2023
internal standard (10 mg/mL) and 100 mL of 2 M NaOH. This mixture was homogenized by vortex
KEYWORDS
(2800 rpm/10 s) and 400 mL of hexane (extractor solvent) with 600 mL of methanol (dispersing solvent) DLLME; antidepressants; LC-
were added to the sample. After the vortex step (2800 rpm/5 s), an ultrasonic bath for 300 s was DAD; whole blood; plasma
employed. Then, this content was centrifuged (10 min/10000 rpm), organic phase was collected and
dried under air flow. After, 30 mL of the mobile phase was used for resuspension and 20 mL is injected
into LC-DAD. This method was optimized and fully validated according to UNODC and SWGTOX guide­
lines, reaching limits of detection equivalent to analytical methodologies that employ mass spectrometry
(MS). Also, it was applied in real cases involving suspected exposure to TCAs. So, the developed DLLME
for the determination of TCAs in whole blood and plasma samples proved to be a simple, reliable,
robust and reproducible method that can be used in toxicology and clinical laboratories.

Introduction concentrations of these molecules are still based on plasma


samples. In addition, hospital laboratories transfer plasma
Medicines are the main agents responsible for the poisoning
samples to measure these xenobiotics in antemortem situa­
in Brazil, responsible for almost 14000 cases registered in the
tions (Perry et al. 1994; Lundstrøm et al. 2022).
country, according to the latest data released by the
Sample handling is an essential step in toxicological analy­
National System of Toxic-Pharmacological Information
ses and dispersive liquid-liquid microextraction (DLLME) has
(SINITOX) (SINITOX (Sistema Nacional de Informaço ~es To
�xico-
been gaining ground in the analyzing of biological samples as

Farmacologicas) 2020). Tricyclic antidepressants (TCAs) are
it is a fast, simple, low-cost technique that can be applied to
one of the main classes of medications involved in cases of
any laboratory. DLLME is based on a mixture of solvents in a
poisoning due to the high number of prescriptions and toxi­
ternary system, which extraction solvent and dispersive solvent
cological complications such as myocardial depression and
are injected into the matrix and quickly the microdrops form
cardiac arrhythmias when administered alone or concomi­
the desired cloud effect. Consequently, a large surface area
tantly with other drugs, which can be lethal (Liebelt and
allows almost instantaneously the analyte extraction (Martins
Francis 2002; SINITOX (Sistema Nacional de Informaço ~es
et al. 2012; Tabani et al. 2019; Manousi and Samanidou 2020).
To�xico-Farmacolo�gicas) 2020).
So, this study aimed to the develop a DLLME procedure
These factors can justify toxicological analysis of TCAs in
for TCAs in whole blood and plasma samples employing
biological samples in order to clarify poisoning cases. Among
liquid chromatography with diode array detector (LC-DAD)
the biological matrices used in toxicological laboratories,
and its application in real cases.
whole blood samples stand out, as highly complex matrix,
easily obtainable and can be used both in clinical emergen­
cies and postmortem cases. Whole blood samples reflect the Experimental
concentrations of TCAs and their respective biological effects,
Chemicals
which make it possible to find precursor drugs as well as
their metabolites for TCAs (Titier et al. 2007; Montenarh et al. Reference standards of amitriptyline (AMI), imipramine (IMI),
2014; De Boeck et al. 2018). However, therapeutic doxepin (DOX), nortriptyline (NOR), desipramine (DES) and

CONTACT Andr�e Valle de Bairros andre.bairros@ufsm.br Nucleous Applied to Toxicology (NAT), Department of Clinical and Toxicological Analysis, Federal
University of Santa Maria, Santa Maria, Brazil.
� 2023 Informa UK Limited, trading as Taylor & Francis Group
2 D. G. BERLATO ET AL.

medazepam (internal standard-IS) were purchased from Whole blood and plasma samples
Sigma-AldrichVR (Saint Louis, USA). Reference standards of
Drug-free whole blood and plasma samples containing 2%
AMI, IMI, DOX, NOR, DES and IS were individually diluted
sodium fluoride preservative (2 mL) from the blood bank of
with methanol (10 mg/mL). The analytes evaluated in this
the Santa Maria University Hospital (HUSM) (Santa Maria,
experiment were prepared to obtain the corresponding
Brazil) were employed for optimization and validation of the
working solutions (1 mg/mL and 100 mg/mL). All stock and
method. Blood samples were collected from suspected
working solutions were stored at −5 � C and kept away from
human intoxication from HUSM. This study was approved by
light. Acetonitrile, methanol, potassium phosphate mono­
the Federal University of Santa Maria Ethics Committee,
basic (KH2PO4), sodium hydroxide (NaOH), phosphoric acid
Brazil (Ethics Protocol Approval N� . 2.353.688).
85% (v/v) and sodium chloride (NaCl) were acquired from
Neon Commercial (Suzano, S~ao Paulo, Brazil).
Chromatography conditions

Instrumentation The stationary phase was WatersV R XBridge C18 3.5 mm


4.6 � 50 mm column and the mobile phase for the present
It was employed an ultrasonic bath Cristo �foliV
R (Campo study was an aqueous solution containing KH2PO4 100 mM
Mour~ao, Brazil), DM22 pHmeter DigimedV R (S~
ao Paulo, Brazil), at pH 2.5 corrected with o-phosphoric acid 85% and metha­
centrifuge 5427 R EppendorfV R (AG, Germany), vortex shaker
nol (60:40), with a flow rate of 1.2 mL/min for the chromato­
BiomixerV R QL-901 and analytical balances AX200 ShimadzuV R graphic run. IS was analyzed at a wavelength of 255 nm;
(Kyoto, Japan). DOX, AMI and NOR at 239 nm; IMI and DES at 249 nm.
For chromatographic apparatus, it was used column
AgilentVR Zorbax Eclipse Plus C18 3.5 mm x 2.1 mm x 50 mm

(Milford, USA) and column ScharlauV R Kromaphase 100 C18


DLLME procedure for sample preparation
3.5 mm x 2.0 mm x 50 mm (Barcelona, Spain) with cartridges In a 2 mL microtube were added 15 mg of NaCl, 490 mL of a
PhenomenexV R C18 4.0 mm x 3.0 mm (Torrance, USA), while
biological sample (whole blood or plasma), 10 mL of IS
TCAs determination occurred in a Nexera - XR - Ultra High- (10 mg/mL) and 100 mL of 2 M NaOH. This mixture was
Performance Liquid Chromatograph (ShimadzuV R, Kyoto, homogenized by vortex (2800 rpm) for 10 s. After this step,
Japan) – equipped whit a 20ADXR pump; SIL-20AC autosam­ the sample was added to a mixture containing 400 mL of
pler system; CTO 10AS Column Oven and SPD-M20A hexane (extractor solvent) with 600 mL of methanol (dispers­
Photodiode Array Detector – Acquisition data from software ing solvent). This mixture was homogenized for 5 s on vor­
LabSolution (ShimadzuV R , Kyoto, Japan). tex (2800 rpm) and bring it to an ultrasonic bath for 300 s.
Then, this content was centrifuged for 10 min at 10000 rpm.
The organic phase was collected, reallocated in a vial, and
Method optimization dried under airflow. In the end, 30 mL of the mobile phase
Optimization was performed taking into consideration the was used for resuspension, and 20 mL was injected into
choice of mobile phase composition based on the conditions LC-DAD. .
found in the literature (Queiroz et al. 1995; Theurillat and
Thormann 1998; Karpinska and Starczewska 2002; Alves et al. Validation of the method
2006; Samanidou et al. 2007; Malfar�a et al. 2007; Uddin et al.
2008; Mercolini et al. 2010) and DLLME parameters. The After DLLME optimization, the method was validated for
DLLME parameters studied were extractor solvent (hexane, whole blood and plasma samples. Validation parameters fol­
toluene; chloroform); essential oil from Eucalyptus (Eucalyptus lowed the United Nations Office on Drugs and Crime valid­
globulus) and orange (Citrus sinensis); dispersing solvent ation guide (UNODC (United Nations Office on Drugs and
(methanol, acetonitrile, isopropanol, ethanol and acetone); Crime) 2009). Thus, the parameters of specificity/selectivity,
vortex effect (0-60s) and ultrasonic bath time (0-300s); selec­ the limit of detection (LoD), the limit of quantification (LoQ),
tion of alkaline sample solution (1-5 M NaOH); centrifugation recovery, linearity, intra- and inter-day precision, accuracy
speed and time (4.000-14.000 rpm, 5-10 min); effect of vol­ and robustness were evaluated (UNODC (United Nations
Office on Drugs and Crime) 2009). To assess both carryover
ume of the extraction and dispersive solvent (300-500 mL;
effect and dilution integrity, the Scientific Working Group for
600-800 mL, respectively). The salting-out effect was also
Forensic Toxicology (SWGTOX) Standard Practices for Method
tested by adding 0, 10 and 15 mg of NaCl (w/v) to the sam­
Validation in Forensic Toxicology guideline was employed
ple before extraction. The stability of the TCAs in extracts
(SWGTOX (Scientific Working Group for Forensic Toxicology)
from the DLLME procedure was evaluated during 24h stor­
2013).
age at −10 � C. Stability tests were performed in triplicate. For
the DLLME optimization procedure, drug-free human sam­
ples have been fortified at a concentration of 200 ng/mL for Limit of detection (LoD) and limit of quantification (LoQ)
each analyte and then submitted to the previously described First, the limit of blank (LoB) was established to ensure a reli­
methods. able LoD. . The LoB is the highest apparent concentration
TOXICOLOGY MECHANISMS AND METHODS 3

Figure 1. Step-by-step of the proposed DLLME.

expected when 60 replicates of 20 blank samples containing time of interest were compared to those from whole blood
no analytes are tested. The data is expressed in the following samples spiked with the analytes at the LoQ. The method
equation: LoB ¼ mean blank þ 1.645 (standard deviation was also evaluated for potential interfering substances
blank). With the LoB previously established, LoD can be through the analysis of free-drug whole blood and plasma
defined as the lowest analyte concentration in a sample that samples spiked with caffeine, acetylsalicylic acid, dipyrone,
can be distinguished from the LoB with significant reliability paracetamol, warfarin, furosemide, atenolol, ibuprofen,
and at which detection is feasible (Armbruster and Pry 2008). phenobarbital, clonazepam, ephedrine, phenylpropanolamine
The procedure to determine LoD was performed similarly and salbutamol at a concentration of 1000 ng/mL.
to that of LoB. Drug-free whole blood and plasma samples Acceptance criteria for this assay were based on the absence
enriched with decreasing concentrations of the analyte of of interfering substances at the retention times of the analy­
interest were used and the results were plotted through the te(s) of interest and their respective IS (UNODC (United
equation: LoD ¼ LoB þ 1.645 (standard deviation low con­
Nations Office on Drugs and Crime) 2009; SWGTOX (Scientific
centration sample). When a value of 1.645 is used as a stand­
Working Group for Forensic Toxicology) 2013).
ard deviation, no more than 5% of the values should be less
than the LoB. If the observed LoD sample values meet this
criteria, then LoD is considered established or verified Linearity
(Armbruster and Pry 2008). Linearity was performed in whole blood and plasma samples
Initially, LoQ was experimentally evaluated in triplicate at in sextuplicate for each point of the analytes of interest,
concentrations of 5, 10, 15, 20 and 30 ng/mL for AMI, DOX starting with the LQ (10 ng/mL for AMI and DOX; 20 ng/mL
and NOR while 5, 10, 15, 20, 30, 40 and 50 ng/mL for IMI and
for NOR and 30 ng/mL for IMI and DES) followed by of 50,
DES. After reaching the acceptance criteria (relative standard
100, 200, 300, 400 and 500 ng/mL. These concentrations
deviation � 20%), a sextuplicate was performed to confirm
were selected according to therapeutic and toxic concentra­
the referred value for each analyte. Acceptance criteria for
tions of TCAs. For acceptability, the linearity must obtain a
the LoQ were coefficient of variation � 20% (relative stand­
correlation coefficient r � 0.99.
ard deviation) for precision and 80-100% for accuracy for
both samples (Armbruster and Pry 2008, UNODC (United
Nations Office on Drugs and Crime) 2009; SWGTOX (Scientific
Recovery
Working Group for Forensic Toxicology) 2013).
Two sets of whole blood and plasma samples were respect­
ively prepared, and set A of samples consisted of the con­
Specificity/selectivity centrations of 45, 250 and 420 ng/mL and it were prepared
Whole blood and plasma samples were extracted and ana­ as described in the topic DLLME procedure for sample prepar­
lyzed according to the previously described procedure for ation for both biological samples. For set B, the analytes
assessing endogenous substances. Peaks at the retention were spiked into the samples immediately after the DLLME
4 D. G. BERLATO ET AL.

step, it was added at the same concentration as set A. The samples were analyzed in sextuplicate. After extraction and
entire procedure was performed in sextuplicate. injection of this sample, three blank samples were injected
Absolute recovery was obtained by comparing the mean to evaluate the memory (carryover) effect.
response obtained for both set A (processed) and the
response for set B (unprocessed). The unprocessed response
Liquid chromatography-mass spectrometry analysis
represented 100% recovery. The UNODC Analytical Methods
Validation Guide (UNODC (United Nations Office on Drugs It was determined by Franco de Oliveira and collaborators
and Crime) 2009) does not consider this factor as indicative (Franco de Oliveira et al. 2019) with Nexera X2 ultra high-
of test failure, as long as the other validation parameters performance liquid chromatography (UHPLC), which con­
achieve the desired purposes. sisted of a degasser, a binary pump, and an autosampler
coupled to an LC-MS 8050 mass spectrometer (ShimadzuV R,

Kyoto, Japan) with an electrospray source operating in the


Precision and accuracy
positive (ESIþ) and negative (ESI–) ion modes, in two separ­
Precision was determined by intra- and inter-day replications.
ate chromatographic runs. The chromatographic separation
This parameter was assessed by analyzing whole blood and
was achieved on a Raptor Biphenyl column (50 mm � 3 mm,
plasma samples containing three known concentrations of
2.7 lm; Restek, USA) eluted with a flow rate of 600 lL/min
target analytes in sextuplicate on three consecutive days. and 45 � C.
Concentrations were 45 ng/mL (lowest level), 250 and For biological samples analysis, an aliquot of 800 lL of an
420 ng/mL. After data collection, a Two-way ANOVA was acetonitrile/methanol mixture (80:20, v/v) was added to the
performed. biological samples (100 lL) spiked with 20 lL of the internal
For accuracy, a curve was developed from their respective standards mix (IS mix; 0.5 lg/mL), and the mixture was
LoQ levels (10 ng/mL for AMI and DOX; 20 ng/mL for NOR shaken for 30 s. After centrifugation for 6 min at 9000 � g, a
and 30 ng/mL for IMI and DES) followed by 200 and 500 ng/ 3 lL aliquot was directly injected into the LC-MS/MS system
mL for all analytes. Afterward, whole blood and plasma sam­ (Franco de Oliveira et al. 2019).
ples fortified with the analytes at concentrations of 45 ng/mL
(lowest point), 250 and 420 ng/mL were evaluated in sextu­
plicate. Accuracy was expressed as a percentage of the Gas chromatography-mass spectrometry (GC-MS)
known concentration, i.e. the acquired mean concentration/ analysis
nominal concentration � 100. Biological samples were treated according to the purposed
DLLME. However, these extracts were resuspended with
Robustness 30 mL of hexane and injected into the equipment. The ana­
Robustness is the ability to resist minor and deliberate varia­ lysis was performed using a ShimadzuV R GC-2010 Plus model

tions in analytical parameters, indicating confidence in the gas chromatography coupled with ShimadzuV R single quadru­

methodology. The following tests were performed: mobile pole mass spectrometry GCMS-Q2010 Ultra model (Kyoto,
phase pH variation (± 0.2 units), employment of different Japan). The samples were injected by splitless mode (1 lL)
chromatographic equipment, laboratories and laboratory into the GC-MS using the ShimadzuV R autosampler AOC-20is

specialists. model (Kyoto, Japan).


Chromatography separation was achieved on a Restek
CorporationV R Rtx-5MS fused-silica capillary column (30 m �

Dilution integrity and carryover effect 0.25 mm i.d. � 0.25 mm thickness film) (Bellefonte, EUA) con­
Dilution integrity is a parameter that allows the evaluation of taining 5% diphenyl and 95% dimethyl polysiloxane. Helium
samples with analyte levels above the calibration curve. was used as carrier gas with 0.8 mL/min at a constant flow
When dilution integrity is required, it should be demon­ rate mode. Column oven temperature program was as fol­
strated by spiking the matrix with analytes above the highest lows: 125 � C (hold 1 min), then programmed at 50 � C/min to
concentration of the curve and then diluting this sample 190 � C; 5 � C/min to 225 � C (hold 3 min); 50 � C/min to 280 � C
with a blank matrix (at least five determinations per dilution (hold 1 min). The total analytical time was 13.40 min. The
factor). Accuracy and precision should be within the set crite­ injection port and transfer line were set at 220 � C and 280 � C,
ria, i.e. within ±15%. respectively. The MS was operated by electron ionization
First, a curve was constructed from respective LoQ (10 ng/ (70 eV) in scan mode. The following ions (m/z) were chosen
mL for AMI and DOX; 20 ng/mL for NOR; 30 ng/mL for IMI to identify the analytes: DOX (178, 115 and 42); IMI (280, 234
and DES), 200 and 500 ng/mL in triplicates. An aliquot of and 220); DES (234, 195 and 193); AMI (58, 91, 202, 189);
500 mL of whole blood and plasma samples were contami­ NOR (263, 128 and 44); IS (207 and 242).
nated with a pool of target analytes with a concentration of
3000 ng/mL. These samples were diluted ten times with
Real case samples application
respective biological matrices to obtain a final concentration
of 300 ng/mL. The IS was added to a final concentration of Whole blood and plasma samples were collected from TCAs
200 ng/mL, and extraction was performed as described in the user volunteers as well as suspected cases of TCAs intoxica­
topic DLLME procedure for sample preparation. Diluted tion from HUSM. The samples from volunteers were collected
TOXICOLOGY MECHANISMS AND METHODS 5

in tubes containing 2% sodium fluoride preservative (2 mL). loss at the end of the process, despite the difficulties in the
In suspected poisoning/exposure by TCAs from the HUSM, formation of the typical cloud effect of DLLME.
available whole blood and plasma samples were used and An interesting alternative is the employment of low-dens­
submitted to the developed methodology. ity solvents (e.g. 1-octanol, 1-nonanol, 1-decanol and other
solvents) as extractor solvents because of their respective
coefficient partition (Log P: � 3.0) and ease of handling.
Results and discussion
However, during our experiments, we observed an increase
Mobile phase constitution and stability in the HPLC system pressure throughout a vast sequence of
injections, making the chromatographic analysis less viable,
The mobile phase that presented the best results for the pre­
as previously described by Saldanha and collaborators
sent study is constituted by an aqueous solution containing
(Saldanha et al. 2022). To avoid future problems in the chro­
100 mM of monobasic potassium phosphate, pH 2.5 cor­
matographic system, we aborted the planned tests with
rected with o-phosphoric acid 85% and methanol in the pro­
these solvents.
portion of 60:40. In this condition, AMI and NOR do not co-
elute, which allowed the simultaneous analysis of these
TCAs. We observed that the mobile phase’s acidification Dispersive solvent
helps stabilize reversed-phase chromatographic columns, The dispersing solvent must be soluble both in the aqueous
resulting in more symmetrical peaks and, consequently, phase and in the organic phase, and it must guarantee the
higher resolution for weak alkaline substances, as described dispersion of the solvent in the aqueous phase, in order to
by Borges (Borges et al. 2012). promote the ‘cloud effect’ required by DLLME (Martins et al.
2012; Tabani et al. 2019; Manousi and Samanidou 2020). The
Optimization of DLLME procedure dispersion formed (cloud effect) is directly linked to the
extraction efficiency, so the choice of this solvent must be
Salting-out effect carefully evaluated. Methanol, acetonitrile, isopropanol, etha­
The salting-out effect promoted an improvement in the nol and acetone were evaluated as dispersing solvents as
extraction because of ionic strength. Our results showed the well as the absence of dispersing solvent for DLLME.
best condition with 15 mg of NaCl (w/v), except for DES. It Acetone and ethanol showed a low rate of enrichment
should be considered that the salting-out effect has variabil­ when observed in the absolute area of the chromatograms.
ities depending on the extraction procedure, matrix sample Indeed, chromatograms generated from the presence of
and specific analytes (Saldanha et al. 2022). Despite of the acetone as a dispersing solvent did not allow the analysis of
particular behavior of DES, 15 mg of NaCl (w/v) was the most DOX due to matrix interferences that coeluted with this ana­
effective for extraction.
lyte. Isopropanol demonstrated a milky appearance of solv­
ent after extraction. So, in order to prevent the integrity of
Extraction solvent the system chromatographic, this specific extract was not
Chloroform, toluene and hexane were evaluated as extractor injected. Acetonitrile and methanol were shown to be effi­
solvents considering their respective partition coefficient cient as dispersing solvents, presenting a low effect of matrix
(Log P: 1.97-3.90). Chloroform did not adequately separate interferents. In this sense, methanol was the best-dispersing
the organic phase from whole blood sample, so it was not solvent among all tested solvents, promoting a cloud effect
injected to preserve the chromatographic column. and, consequently, a higher absolute area of the
Considering the characteristics present in biological samples, chromatograms.
especially whole blood samples (high concentration of cells
and consequently high molecular weight endogenous mole­
cules) (Skopp 2004; Peliç~ao et al. 2018), solvents with a dens­ Effect of vortex and ultrasonic bath time
ity greater than water (e.g. chloroform) could cause Ultrasonic bath and vortex have been used in several studies
difficulties when removing the organic solvent from complex to reduce solvent amounts and increase DLLME extraction
biological samples such as whole blood samples as well as efficiency (Fern�andez et al. 2016; Vaghar-Lahijani et al. 2018;
the amount of interferences from biological matrix extracted Fern�andez et al. 2019; Carasek et al. 2021; Tomai et al. 2021;
by the solvent. Saldanha et al. 2022). The extraction efficiency with vortex
As an extracting solvent, toluene could extract the TCAs homogenization at 2800 rpm was evaluated at times of 0, 15,
in whole blood samples (Chen et al. 2017). However, we 30 and 60 s. Vortex homogenization for 30 s showed the best
observed the significant of endogenous interferences as well result and this tested condition was compared with ultra­
as waste of solvent volume during the DLLME, similar to sound bath in 30, 60, 150 and 300 s with 40 KHz frequency at
Saldanha e coauthors’ description (Saldanha et al. 2022). 35 � C. Before the ultrasonic bath assay, vortex homogeniza­
Hexane was the solvent with the higher higher extraction tion at 2800 rpm for 5 s was used in order to maximize the
capacity as expected, considering its high partition coeffi­ extraction procedure that is promoted by the ultrasound
cient (Log P: 3.90), a close value to the TCAs evaluated (Log bath. An ultrasonic bath for 300 s to homogenize the dispers­
P: 3.84-4.92). In addition, hexane showed less endogenous ing and extractor solvents in whole blood sample was better
interferences among the tested solvents with little volume than vortex homogenization.
6 D. G. BERLATO ET AL.

Selection of the pH of sample solution approach is employed successfully for liquid-liquid extraction
Because of matrix interference, visualization of the pH value (Meatherall 1994; Fu et al. 2010) and hollow-fiber liquid-
of the biological matrix was not able to be performed, so we phase microextraction (LPME) (De Bairros et al. 2015). So, a
opted for fixed volume in different concentrations of NaOH. mixture containing hexane:1-octanol was evaluated in the
In this experiment, the extraction efficiency based on pH cor­ following proportions: 95:5, 90:10 and 85:15.
rection was evaluated from 1, 2 and 3 M NaOH. The addition However, this approach did not obtain satisfactory results,
of 100 mL of 2 M NaOH showed greater extraction effective­ in spite of the partition coefficient of 1-octanol (Log P: 3.0).
ness compared to other NaOH concentrations. Non-addition The drying process of 1-octanol, even in a low volume,
of the pH corrector resulted in the non-extraction of TCAs required a time of more than 30 min for total drying. In add­
from the whole blood sample by the purpose of DLLME. ition, it was not efficient in the extraction of TCAs in whole
Blood samples pH ranges from 7.35 − 7.45 in antemortem blood samples because of the amount of matrix interferents,
situations; however, blood matrices from postmortem cases which precluded the chromatographic analysis of DOX and
tend to acidify due to the various processes that occur after IS. Such behavior prevented carrying out the chromato­
death (Donaldson and Lamont 2013; Peliç~ao et al. 2018). In graphic evaluation of the analytes in question for the opti­
both situations, TCAs present in the blood matrices are in mization step (data not shown).
the ionized form due to their respective pKa values (9.4 to
10.02), so the pH must be adjusted to at least 2 points above
the pKa of the molecules (pH > 12.02) to guarantee in non- Stability study from DLLME procedure
ionized form for the mass transfer of said substance from the Freshly prepared samples from the DLLME procedure were
biological matrix to the organic solvent to occur (Flanagan measured and compared with prepared samples in the day
et al. 2007; Dos Santos et al. 2014). before and stored at −10 � C in a capped vial. There was a
significant loss of analytes, except DOX, in the storage pro­
cess for one day in the conditions tested. The results are
Centrifugation speed and time expressed in the amount of analyte lost in percentage (%)
Centrifugation speed and time were evaluated in the follow­ and it was found 5.5, 23.1, 33.33, 60 and 81% for DOX, IMI,
ing conditions: 10 min at 4.000 rpm, 10 min at 10.000 rpm and DES, AMI and NOR, respectively. Probably the oxygen atom
5 min at 14.000 rpm. The best condition was 10 min at present in the tricyclic ring from DOX allowed greater stabil­
10.000 rpm showed less matrix interference when centrifuged ity of the analyte under the imposed conditions compared
for 10 min at 4.000 rpm and there was no difference compared to the other TCAs. However, studies to assess the stability of
to centrifuging for 5 min at 14.000 rpm. In addition, during extracts from DLLME for TCAs are needed to confirm this
this optimization step, there was a risk of rupture of the flasks theory.
used with the centrifugation force at 14000 rpm. During this Traditional freeze-thaw cycle tests assessing the stability
experiment, we observed that centrifugation helped to break of these analytes in biological samples have already been
the emulsion formed by DLLME in whole blood samples, related in the literature (Seidi et al. 2013; Mohebbi et al.
becoming a vital step for this type of extraction, as described 2018b). However, our objective was the stability of the
by Saldanha and coauthors (Saldanha et al. 2022). extract from DLLME to support an experiment logistics pro­
posal, which unfortunately did not prove to be viable. So, for
Effect of volume on extractor and dispersing solvent validation of the method, the entire stage of the developed
Different volumes of the extractor and dispersing solvents method was prepared and evaluated on the same day.
(300, 400 and 500 mL; 600, 700 and 800 mL, respectively) were
evaluated. Increasing the volume of the extractor solvent with
Essential oils as extractor solvent
subsequent may result in a lower pre-concentration factor of
Based on the principles of Green Chemistry, we evaluated
the analytes due to the appropriate ratio between the organic/
essential oil from Eucalyptus (Eucalyptus globulus) and orange
aqueous phase (matrix). In order to maximize DLLME extrac­
(Citrus sinensis) in place of the hexane solvent because of
tion, dispersing solvent volume is studied to reach a constant
their drying and extraction capacity as described by Silveira
volume of the organic phase. So, the appropriate volume of
and coauthors (Silveira et al. 2021).
dispersing solvent for microdroplet formation depends on
We found some difficulties in promoting the TCAs extrac­
both the volume of the aqueous phase and the volume of the
tion with this methodology. In spite of methanol showing
extractor solvent (Martins et al. 2012; Carasek et al. 2021;
some potential as a dispersing solvent while essential oil
Saldanha et al. 2022). In this sense, the best condition was
from Eucalyptus (Eucalyptus globulus) is employed as an
respectively 400 and 600 mL of extractor and dispersing solvent
extractor solvent, this approach did not demonstrate total
while lower volumes showed no increase in DLLME capacity.
compatibility considering the proposed DLLME technique.
For essential oil from orange (Citrus sinensis), dispersing
Mixture of extracting solvents for DLLME solvent did not promote cloud effect as DLLME purpose.
In an attempt to increase the extraction efficiency, a mixture Therefore, the low volume available after extraction and han­
of extracting solvents was performed considering the parti­ dling of this essential oil presented difficulties. In addition,
tion coefficient of TCAs as well as extractor solvents. This both essential oils used in the proposed DLLME technique
TOXICOLOGY MECHANISMS AND METHODS 7

demonstrated several matrix interferents, which it did not establishing specificity/selective, linearity, recovery, intra and
allow the evaluation of analytes in whole blood samples. inter-day precision, accuracy, limit of detection (LoD), limit of
In an attempt to reduce the interference of whole blood quantification (LoQ), robustness, dilution integrity and carry­
samples in the extraction with essential oils for the DLLME over effect. No interfering peaks due to endogenous interfer­
technique, a pretreatment step was performed with 10% tri­ ents or exogenous substances (mainly general and prescribed
chloroacetic acid and 50% NaOH. Despite being able to sep­ drugs used concomitantly with TCAs) were observed at the
arate the phases and facilitate the collection of essential oil, retention time of the compounds of interest.
this strategy was not effective in extracting the TCAs. Considering the equipment used, extraction technique
Perhaps strong acid and alkaline solutions can lead to the and the complexity of the matrices of choice (whole blood
hydrolysis of the analytes, making the quantification of TCAs and plasma samples), LoD values as well as the LoD, were
unfeasible. Therefore, the pretreatment step did not elimin­ satisfactory within the therapeutic range (2.5-10 and 10-
ate the matrix interferents from the biological sample. A 30 ng/mL, respectively). In fact, the low levels of LoD and
summary of the data from all optimized parameters can be LoQ reached according to this method correspond to con­
seen in Figure 2. centrations found in methodologies that employ liquid chro­
matography or gas chromatography coupled to mass
spectrometry (LC-MS or GC-MS) (Titier et al. 2007; Dos Santos
Validation of the method
et al. 2014; Santos et al. 2017).
Based on data from whole blood method optimization in The linearity parameter was established from their
this study and greater complexity of whole blood samples respective LoQ up to 500 ng/mL, which the phenomenon of
compared to plasma samples for the determination of xeno­ heteroscedasticity, evaluated through the distribution F
biotics, it was considered the possibility of measurement of (Almeida et al. 2002), was not observed. So, homoscedasticity
TCAs in plasma samples for antemortem situations as well as behavior was found in all analytes. Therefore, it was not
the therapeutic concentrations of TCAs are based on the needed weighted squares linear regression. The respective
respective matrix (Perry et al. 1994; Titier et al. 2007; calibration curves and coefficients of correlations for whole
Montenarh et al. 2014; De Boeck et al. 2018; Lundstrøm et al. blood samples were y ¼ 0.0038x − 0.0314; r ¼ 0,9961 (DOX);
2022). So, validation of the method was applied directly to y ¼ 0.0006x − 0.0034; r ¼ 0.9976 (IMI); y ¼ 0.0012x þ 0.0064;
plasma samples without the optimization step. r ¼ 0.9967 (DES); y ¼ 0.0044x − 0.0553; r ¼ 0.9928 (AMI) and
Validation of the method was conducted following UNODC y ¼ 0.0035x − 0.0237; r ¼ 0.9933 (NOR). For plasma samples,
and SWGTOX guidelines (UNODC (United Nations Office on it were found y ¼ 0.0036x − 0.0383, r ¼ 0.9973 (DOX);
Drugs and Crime) 2009; SWGTOX (Scientific Working Group y ¼ 0.0007x − 0.0136, r ¼ 0.9987 (IMI); y ¼ 0.0021x − 0.0394,
for Forensic Toxicology) 2013) after optimization of the DLLME r ¼ 0.9947 (DES); y ¼ 0.0043x − 0.0204, r ¼ 0.9982 (AMI);
procedure of these analytes and was carried out by y ¼ 0.006x − 0.1218, r ¼ 0.9966 (NOR).

Figure 2. Optimization of DLLME for TCAs in biological samples. AMI, amitriptyline; DES, desipramine; DOX, doxepine; IMI, imipramine; nor, nortriptyline. UB, ultra­
sonic bath. Vortex with 2800 rpm and ultrasonic bath with 40 KHz frequency at 35 � C.
8 D. G. BERLATO ET AL.

Regarding the intra and inter-day precision, the values for robustness, with satisfactory results, and can be used in
these analytes at three different concentration levels were other laboratories.
less than 15% in all concentration levels. These data were Furthermore, this methodology can be reproduced in GC-
considered suitable for all tested substances for intra and MS, which the final extract was resuspended with hexane for
inter-day precision (RSD% < 12.80 and 14.59, respectively). posterior injection, allowing the TCAs confirmation. However,
Accuracy values of the TCAs evaluated at three different con­ this approach has not been properly validated by following a
centration levels ranged from 85.52 to 113.67% for whole validation guideline. The validation parameters of the pro­
blood samples while plasma samples ranged from 91.83 to posed method are summarized in Table 1.
114.44%. In both validation parameters, the values found are DLLME studies reported in the literature in urine samples
within the criteria required by the UNODC and SWGTOX val­ showed 3-8 ng/mL as LoD for AMI, clomipramine (CLO) and
idation guides (UNODC (United Nations Office on Drugs and thioridazine (TIO) (Xiong et al. 2009); 0.6 ng/mL for IMI and
Crime) 2009; SWGTOX (Scientific Working Group for Forensic trimipramine (TRI) (Shamsipur and Mirmohammadi 2014);
Toxicology) 2013). 240-310 ng/mL for AMI, DES, IMI, NOR and CLO (Mohebbi
Recovery rates showed variations (9.86% to 31.60%; 29.20 et al. 2018) employing liquid chromatography with ultraviolet
to 51.49% for whole blood and plasma samples, respect­ detector (LC-UV) while gas chromatography with flame ion­
ively), considering the simplicity of the extractive technique ization detector (GC-FID) detected 2 ng/mL for NOR (Nabil
used, the complexity of the biological matrices and the other et al. 2015). For plasma samples, Bazregar and collaborators
results of the validation parameters, the recovery rates for (Bazregar et al. 2016) achieved 0.7-1.0 ng/mL for AMI, IMI and
the TCAs are according to the literature (Dos Santos et al. NOR using LC-UV while Vaghar-Lahijania and coauthors
2014; Vaghar-Lahijani et al. 2018; Dos Mohebbi et al. 2018). (Vaghar-Lahijani et al. 2018) determined 0.5 ng/mL for AMI
Whole blood samples showed more complexity than plasma employing LC-DAD. Yazdi and coauthors (Yazdi et al. 2008)
samples, and it could be observed in the difficulty in forming detected 2 ng/mL of NOR through GC-FID.
the DLLME cloud effect. Consequently, low recovery was Considering DLLME studies with the employment of gas
obtained in the proposed method. However, this parameter chromatography coupled to mass spectrometry (GC-MS), Ito
is not considered indicative of test failure as long as the and collaborators (Ito et al. 2011) detected 0.5 − 2 ng/mL in
other validation parameters achieve the desired purposes urine samples for AMI, CLO, DES IMI and NOR while concen­
(UNODC (United Nations Office on Drugs and Crime) 2009; trations between 0.013 − 0.025 ng/mL for AMI, DES and CLO
SWGTOX (Scientific Working Group for Forensic Toxicology) were considered as LoD by Mohebbi and coauthors
2013). (Mohebbi et al. 2019). Liquid chromatography coupled to
For intoxication cases, it is not uncommon to encounter tandem mass spectrometry (LC-MS/MS) was used to deter­
high TCAs concentrations, possibly above the highest point mine CLO, IMI and other psychoactive substances which
of the studied calibration curve (500 ng/mL). So, the inclusion 2 ng/mL was identified as LoD for TCAs in whole blood sam­
of the dilution integrity parameter at known levels (e.g.: 1:10; ples (Fisichella et al. 2015). In fact, Fisichella and coauthors
1:5) is important to ensure reliable results. This parameter study (Fisichella et al. 2015) is the only published manuscript
was performed by spiking the blank whole blood and plasma that used whole blood samples as biological matrix of
samples with 3000 ng/mL and diluting them ten times using choice.
the same matrix (six replicates per dilution factor). Whole Considering studies with full validation from DLLME
blood samples showed 87.30–114.05% for accuracy and 6.93- involving TCAs in biological samples, only Fisichella and col­
9.63% for precision, while plasma samples demonstrated laborators (Fisichella et al. 2015) performed a method with
98.87-114.01% and 2.45-4.30% for accuracy and precision, these characteristics. However, their study used five repli­
respectively. These results are in concordance with the crite­ cates for each concentration for intra and inter-day precision
ria established by SWGTOX (RSD% � 15) (SWGTOX (Scientific as well as accuracy, while three replicates were employed for
Working Group for Forensic Toxicology) 2013). Also, it was the linearity parameter. Therefore, this method did not
observed no memory effect (carryover) in the chromato­ report the phenomenon of heteroscedasticity as well as dilu­
graphic runs. tion integrity and carryover effect for the target analytes.
The robustness parameter is a measure of the method’s Nevertheless, ANOVA was not used to evaluate intra and
capacity to remain unaffected by small and deliberate varia­ inter-day precision as statistical proof. Also, this study
tions. The risk of finding out that a given method does not employed LC-MS as equipment to determine TCAs (Fisichella
fulfill these criteria late in the validation process may result et al. 2015), a much more expensive and rarer instrument in
in the need for it to be redeveloped and optimized (Vander laboratories compared to the LC-DAD. Other studies reported
Heyden et al. 2001). So, it was evaluated variation of the pH in this manuscript did not carry out an adequate full valid­
of the mobile phase (0.2 units) was not relevant to change ation and/or appropriate statistical analysis in biological sam­
the method as well as chromatographic columns from differ­ ples. A summary of the procedures that have used DLLME
ent suppliers (AgilentV R and ScharlauV R ). Also, it was observed for the analysis of TCAs in biological samples available in the
the analysis of TCAs in different equipment (N ¼ 3) and by scientific literature can be seen in Table 2.
laboratory technicians (N ¼ 4) as well as laboratory structures It should be considered that the purposed method is the
(N ¼ 3), and the differences were not relevant. Even with the only one that evaluates DOX among the other procedures
different conditions, the purpose technique showed highlighted in this manuscript. Also, this methodology can
TOXICOLOGY MECHANISMS AND METHODS 9

Table 1. Validation parameters of the developed method for the determination of TCAs in whole blood and plasma samples (six replicates for each
concentration).
Whole blood Plasma
Parameters
DOX IMI DES AMI NOR DOX IMI DES AMI NOR
Recovery (%)
C1 15.66 14.15 14.42 14.45 18.38 48.44 50.46 31.71 52.80 34.31
C2 11.71 13.40 9.86 13.77 11.32 38.99 44.45 23.20 45.73 29.20
C3 21.60 23.30 17.22 23.99 31.60 42.76 51.49 28.98 51.11 33.96
LD (ng/mL) 2.50 10.00 10.00 2.50 7.50 2.50 10.00 10.00 2.50 7.50
LQ (ng/mL) 10.00 30.00 30.00 10.00 20.00 10.00 30.00 30.00 10.00 20.00
Intra-day Precision (CV%)
C1 10.90 12.80 12.03 12.35 9.83 7.46 11.49 9.84 7.32 8.11
C2 8.59 11.20 7.70 9.78 6.71 7.28 10.66 9.44 7.11 9.11
C3 9.53 8.31 9.98 4.46 8.91 4.29 9.22 9.03 10.13 6.45
Inter-day Precision (CV%)
C1 11.90 13.86 13.06 9.67 5.82 7.84 10.47 8.02 9.15 7.01
C2 9.20 7.56 14.59 10.89 11.87 5.57 9.10 5.67 5.00 5.32
C3 11.62 10.18 12.99 7.23 13.40 8.10 11.10 11.28 11.32 9.43
Accuracy (%)
C1 92.89 85.64 113.78 96.87 113.67 113.33 114.44 94.53 113.40 99.40
C2 100.28 93.45 89.62 85.52 89.95 94.86 95.5 96.86 91.83 93.83
C3 111.16 95.92 97.54 90.19 99.18 100.24 99.78 101.38 98.10 98.33
Dilution integrity
Precision (%)
10 times 8.38 6.93 7.87 9.63 8.42 2.68 4.30 2.60 2.45 3.30
Accuracy (%)
10 times (300 ng/mL) 100.67 87.30 114.05 88.99 105.40 114.01 103.64 106.13 98.87 104.28
AMI, amitriptyline; DES, desipramine; DOX, doxepine; IMI, imipramine; NOR, nortriptyline. C1, 45 ng/mL (lowest point); C2, 250 ng/mL; C3, 420 ng/ml. LoD, limit
of detection; LoQ, limit of quantification. RSD%, relative standard deviation em percentage. 10 times, 3000 ng/mL.

Table 2. Summary of DLLME techniques for the determination of TCAs in biological samples.
Detectability
Matrix Analytes Equipment (ng/mL) Extraction procedure Validation parameters Reference
Whole blood and AMI, DES, DOX, IMI LC-DAD 2.50 − 10 UA-DLLME LoD, LoQ, specificity/selective, Purposed method
plasma and NOR linearity, recovery, intra
and inter-day precision,
accuracy, robustness, dilute
integrity and carryover
effect
Urine AMI, DES and CLO GC-MS 0.013 − 0.025 SSI/DES-DLLME-SFO LoD, LoQ, linearity, selectivity, Mohebbi et al. 2019
accuracy, intra and inter-
day precision, stability,
robustness and extraction
recovery
Plasma AMI and LC-DAD 0.5 UA − IL − DLLME LoD, linearity and recovery Vaghar-Lahijani
psychoactive et al. 2018
drug
Urine AMI, DES,IMI, NOR LC-UV 240 - 310 SI-HLLE followed by LoD, LoQ, specificity/selective, Mohebbi et al. 2018
and CLO DSPE and later linearity, recovery, intra-
DLLME-SFO day precision and accuracy
Plasma AMI, IMI and NOR LC-UV 0.7 − 1.0 Traditional DLLME LoD, linearity, intra and inter- Bazregar et al. 2016
day precision
Urine NOR and GC-FID 2.0 DLLME with in situ LoD, LoQ, linearity, intra and Nabil et al. 2015
psychoactive derivatization inter-day precision and
drugs accuracy
Whole blood CLO, IMI and other LC-MS/MS 2.0 Traditional DLLME LoD, LoQ, specificity/selective, Fisichella et al. 2015
psychoactive linearity, recovery, intra-
drugs day precision and accuracy
Urine IMI and TRI LC-UV 0.6 Traditional DLLME LoD, linearity and intra-day Shamsipur and
precision Mirmohammadi
2014
Urine AMI, CLO, DES, IMI GC-MS 0.5 − 2.0 DLLME with in situ LoD, LoQ, specificity/selective, Ito et al. 2011
and NOR derivatization linearity, intra-day
precision and accuracy
Urine AMI, CLO and TIO LC-UV 3.0 − 8.0 DLLME followed by LoD, LoQ, linearity, intra and Xiong et al. 2009
membrane inter-day precision and
filtration 0.45 lm accuracy
Water and plasma AMI and NOR GC-FID 5.0 – 10 Traditional DLLME LoD, linearity, intra-day Yazdi et al. 2008
precision and recovery
AMI, amitriptyline; CLO, clomipramine; DES, desipramine; IMI, imipramine; TRI, trimipramine; TIO, thioridazine. LoD, Limit of detection; LoQ, Limit of quantifica­
tion. DLLME, dispersive liquid phase microextraction; DLLME-SFO, dispersive liquid-liquid microextraction based on the solidification of floating organic droplet;
DSPE, dispersive solid phase extraction; SSI/DES-DLLME-SFO, sodium sulfate-induced deep eutectic solvent-based solidification of floating organic droplets–dis­
persive liquid phase microextraction; SI-HLLE, salt induced-homogenous liquid-liquid extraction; UA − IL − DLLME, ultrasound-assisted ionic liquid-based dispersive
liquid–liquid microextraction. GC-FID, gas chromatography with flame ionization detector; GC-MS, gas chromatography coupled mass spectrometry; LC-UV, liquid
chromatography with ultraviolet detector; LC-DAD, liquid chromatography with diode array detector.
10 D. G. BERLATO ET AL.

analyze TCAs and their respective metabolites in whole Real case 3


blood and plasma samples at low concentrations similar to Postmortem whole blood sample with the suspected pres­
those achieved by GC-MS and LC-MS (Ito et al. 2011; ence of TCAs was submitted to the proposed DLLME. It was
Mohebbi et al. 2019; Fisichella et al. 2015) employing a LC- detected IMI and AMI as demonstrated in Figure 3E.
DAD. Therefore, this is the first method with full validation However, below their LQ (30 and 10 ng/mL, respectively).
following international validation guidelines (UNODC and
SWGTOX) (UNODC (United Nations Office on Drugs and
Crime) 2009; SWGTOX (Scientific Working Group for Forensic
Real case 4
A plasma sample from a patient from HUSM with suspected
Toxicology) 2013) using DLLME in whole blood samples for
drug poisoning from different classes of medicines, including
TCAs employing LC-DAD.
exposure to TCAs, was referred to the laboratory. AMI was
Whole blood samples for toxicological analysis allow
detected in the plasma sample, however, below the LQ
greater versatility for drug determination, as it is feasible
(10 ng/mL). An aliquot of the same plasma sample was ana­
to apply a method to antemortem and postmortem sam­
lyzed in GC-MS using the DLLME proposal in order to iden­
ples (Peliç~ao et al. 2018). Therefore, Amitai and collabora­
tify the analyte. Chromatographic conditions and respective
tors (Amitai et al. 1993) described a red blood cells/plasma
chromatograms are described in Figure 3H and J.
ratio for TCAs, which AMI and IMI showed < 1, DOX is
close to 1 and the other are > 1 for this ratio, respectively.
Similarly, Montenarh and coauthors (Montenarh et al. Real case discussion
2014) found higher values of AMI in whole blood, while
DES showed lower levels of recovery when compared to In real case 1, it was measured AMI and NOR (68.45 ng/mL
serum/plasma. and 31.60 ng/mL, respectively) in whole blood sample. The
Based on this information, this indicates that whole patient arrived at the hospital, and the whole blood sample
blood samples should be the preferred biological matrix for was collected for toxicological analysis 24 h after hospital
the determination of TCAs (Amitai et al. 1993; Titier et al. intervention. The patient admitted the consumption of AMI
2007; Montenarh et al. 2014; De Boeck et al. 2018). as a psychotherapy drug according to medical prescription.
However, therapeutic levels of these substances are still However, the patient did not describe the posology of AMI.
based on plasma samples as well as clinical situations such According to the related from real case 2, the dose of
as hospital laboratories employ this biological matrix to AMI (12.5 mg/day) is used for the beginning of the temporo­
determine TCAs in antemortem cases (Perry et al. 1994; mandibular disorder treatment. This is considered a lower
Lundstrøm et al. 2022). Thus, the purpose methodology was dose than that employed at the end of this treatment (75-
validated with success for plasma samples and can be used 100 mg/day) and even smaller than the posology for depres­
for the determination of these analytes in this biological sion treatment (25- 300 mg/day) (Diaz et al. 2020). The whole
fluid. So, an analytical methodology with full validation for blood sample was collected from the volunteer 1 h after the
both samples improves the capacity of TCAs analysis as medication was taken. According to a study by Garland
well as diversifies the scope of matrices in suspected expos­ (Garland 1977), oral ingestion of 50 mg of AMI in volunteers
ition for these molecules. resulted in a peak plasma concentration of 20-40 ng/mL after
2-4 h while, while Edelbroek and collaborators (Edelbroek
et al. 1984) determined serum level between 50-250 ng/mL
Real case applications after ingestion of 150 mg of AMI. So, the low level of AMI
found in the whole blood sample (19.33 ng/mL) can be
Real case 1
explained by due to the low dosage adopted by the volun­
Whole blood sample from a 17-year-old female was sent to
teer and the period between AMI ingestion and sample
the laboratory. The patient arrived at the hospital with sus­
collection.
pected drug poisoning. An immunochromatographic test
In the real case 3, postmortem whole blood sample with
was performed, and it was detected methamphetamine and
the suspected presence of TCAs was submitted to the pro­
TCAs. The purpose methodology determined AMI and NOR
posed DLLME. There is no information in this specific case
in whole blood samples with 68.45 ng/mL and 31.60 ng/mL,
on the consumption of TCAs, sample collection or storage
respectively (Figure 3C).
time, except the toxicological analysis performed by LC-MS.
However, the proposed methodology was able to detect IMI
Real case 2 and AMI.
A 23-year-old female was a volunteer to participate in this Real case 4 demonstrated that AMI concentration was
research. The volunteer informed that she uses the dosage below the LQ (10 ng/mL). For toxicological analysis, plasma
of 12.5 mg of AMI and 50 mg of alprazolam to treat temporo­ sample was collected 48 h after hospitalization patient, and
mandibular disorder. Every morning, she takes her medica­ the immunochromatographic test showed negative results
tion during breakfast. The whole blood sample was for all drugs, including TCAs. However, the patient’s family
collected, stored at 4 � C for 24 h, and then submitted to the admitted her consumption of TCAs for depression treatment.
proposed methodology. It was found 19.33 ng/mL of AMI, as Based on the related, toxicological analysis was performed.
demonstrated in the Figure 3D. However, considering the delay between hospital admission
TOXICOLOGY MECHANISMS AND METHODS 11

Figure 3. Chromatograms were obtained by the DLLME and LC-DAD and GC-MS analysis of whole blood and plasma samples. AMI, amitriptyline (RT: 13.13 min);
DES, desipramine (RT: 11.05 min); DOX, doxepine (RT: 4.85 min); IMI, imipramine (RT: 9.83 min); is, internal standard (RT: 3.53 min); nor, nortriptyline (RT: 14.55 min).
RT, retention time. A-H, LC-DAD analysis; I-J, GC-MS analysis. (A) Blank whole blood sample spiked with 200 ng/mL of is. (B) Blank whole blood sample spiked with
200 ng/mL of is while 100, 200 and 400 ng/mL for target analytes. (C) Real sample (case 1) containing AMI and NOR in whole blood sample with 68.45 ng/mL and
31.60 ng/mL, respectively. (D) Real sample (case 2) containing 19.33 ng/mL of AMI in whole blood sample. (E) Real sample (case 3) detected IMI and AMI, however,
below respective LQs in postmortem whole blood sample. (F) Blank plasma sample spiked with 200 ng/mL of is. (G) Blank plasma sample spiked with 200 ng/mL of
is while 100, 200 and 400 ng/mL for target analytes. (H) Real sample (case 4) detected AMI, however, below LQ in the plasma sample. (I) Blank plasma sample
spiked with 200 ng/mL of AMI (retention time: 11.32 min). (J) Real sample (case 4) detected AMI (retention time: 11.32 min).

and sample collection, it was not possible to determine blood samples showed lower viscosity when compared to
whether the patient was intoxicated with TCAs, in spite of antemortem samples, which allowed a more prominent
the detection of AMI in plasma samples. cloud effect characteristic of DLLME. Consequently, extraction
During the development of the proposed DLLME method­ capacity is higher. Only Fisichella and collaborators (Fisichella
ology, it was observed difficulty in the dispersion of solvents et al. 2015) evaluated whole blood samples (50 real cases)
in the whole blood sample because of the greater complex­ from postmortem cases employing DLLME and LC-MS.
ity of this biological matrix if compared to plasma samples. Considering the high capacity of TCAs to bind to plasma
Application of the DLLME method demonstrated different proteins and erythrocytes, the physiological changes caused
behavior between antemortem and postmortem whole by the postmortem phenomenon can release these drugs as
blood for TCAs extraction. Experimentally, postmortem whole well as decrease the viscosity of whole blood samples.
12 D. G. BERLATO ET AL.

However, further studies should be performed to confirm alternativas para melhorar as separaço ~es. Qu�ım Nova. 35(5):993–1003.
this hypothesis. Until this moment, this is the first DLLME doi: 10.1590/S0100-40422012000500024.
Carasek E, Bernardi G, Morelli D, Merib J. 2021. Sustainable green sol­
study involving antemortem and postmortem whole blood vents for microextraction techniques: recent developments and appli­
samples. Also, the proposed DLLME methodology was fully cations. J Chromatogr A. 1640:461944. doi: 10.1016/j.chroma.2021.
validated, and it is able to determine TCAs in whole blood 461944.
and plasma samples with success for toxicological analysis. Chen X, Zheng S, Le J, Qian Z, Zhang R, Hong Z, Chai Y. 2017.
Ultrasound-assisted low-density solvent dispersive liquid-liquid micro­
extraction for the simultaneous determination of 12 new antidepres­
Conclusion sants and 2 antipsychotics in whole blood by gas chromatography-
mass spectrometry. J Pharm Biomed Anal. 142:19–27. doi: 10.1016/j.
Proposed DLLME methodology was able to determine TCAs jpba.2017.04.032.
in whole blood and plasma samples using LC-DAD, and it De Bairros AV, de Almeida RM, Pantale~ao L, Barcellos T, e Silva SM,
Yonamine M. 2015. Determination of low levels of benzodiazepines
proved to be a simple, reliable, robust, and reproducible and their metabolites in urine by hollow-fiber liquid-phase microex­
method that can be used in toxicology laboratories. Thus, an traction (LPME) and gas chromatography-mass spectrometry (GC-MS).
analytical approach with full validation for both samples J Chromatogr B Analyt Technol Biomed Life Sci. 975:24–33. doi: 10.
improves the capacity of TCAs analysis as well as diversifies 1016/j.jchromb.2014.10.040.
De Boeck M, Dehaen W, Tytgat J, Cuypers E. 2018. Ionic liquid-based
the scope of matrices in suspected exposition for these mol­
liquid-liquid microextraction for benzodiazepine analysis in postmor­
ecules. Considering the possibilities of DLLME and TCAs tem blood samples. J Forensic Sci. 63(6):1875–1879. doi: 10.1111/
chemical aspects, further studies should be conducted to val­ 1556-4029.13778.
idate the technique in other biological matrices and chroma­ Diaz D, Vallejos A, Torres S, Hern�andez W, Calvache J, Merch�an J, Latorre
tographic equipment. G, Maldonado L. 2020. Detection of potential risks in the prescription
of tricyclic antidepressants through an online clinical alert system.
Rev Colomb Psiquiat. 49(1):9–14.
Donaldson AE, Lamont IL. 2013. Biochemistry changes that occur after
Acknowledgment
death: potential markers for determining post-mortem interval. PLoS
The authors would like to acknowledge the financial support from the One. 8(11):e82011. doi: 10.1371/journal.pone.0082011.
Coordination of Improvement of Personal Higher Education – Brazil Dos Santos MF, Ferri CC, Seulin SC, Leyton V, Pasqualucci CAG, Mun ~oz
(CAPES – Finance Code 001) for the scholarship provided to Dener DR, Yonamine M. 2014. Determination of antidepressants in whole
Gomes Berlato and Geovane de Almeida Saldanha. blood using hollow-fiber liquid-phase microextraction and gas chro­
matography–mass spectrometry. Forensic Toxicol. 32(2):214–224. doi:
10.1007/s11419-014-0226-9.
Disclosure statement Edelbroek PM, Zitman FG, Schreuder JN, Rooymans HG, de Wolff FA.
1984. Amitriptyline metabolism in relation to antidepressive effect.
The authors declare that they have no known competing financial inter­ Clin Pharmacol Ther. 35(4):467–473. doi: 10.1038/clpt.1984.61.
ests or personal relationships that could have appeared to influence the Fern�andez P, Regenjo M, Ares A, Fern�andez AM, Lorenzo RA, Carro AM.
work reported in this paper. 2019. Simultaneous determination of 20 drugs of abuse in oral fluid
using ultrasound-assisted dispersive liquid-liquid microextraction. Anal
Bioanal Chem. 411(1):193–203. doi: 10.1007/s00216-018-1428-5.
Funding Fern�andez P, Taboada V, Regenjo M, Morales L, Alvarez I, Carro AM,
Lorenzo RA. 2016. Optimization of ultrasound assisted dispersive
Coordenaç~ao de Aperfeiçoamento de Pessoal de N�ıvel Superior. liquid-liquid microextraction of six antidepressants in human plasma
using experimental design. J Pharm Biomed Anal. 124:189–197. doi:
10.1016/j.jpba.2016.02.041.
References Fisichella M, Odoardi S, Strano-Rossi S. 2015. High-throughput dispersive
liquid/liquid microextraction (DLLME) method for the rapid determin­
Almeida AM, Castel-Branco MM, Falc~ao AC. 2002. Linear regression for
ation of drugs of abuse, benzodiazepines and other psychotropic
calibration lines revisited: weighting schemes for bioanalytical meth­
medications in blood samples by liquid chromatography-tandem
ods. J Chromatogr B Analyt Technol Biomed Life Sci. 774(2):215–222.
mass spectrometry (LC-MS/MS) and application to forensic cases.
doi: 10.1016/s1570-0232(02)00244-1. Microchem J. 123:33–41. doi: 10.1016/j.microc.2015.05.009.
Alves C, Fernandes C, Dos Santos Neto AJ, Rodrigues JC, Costa Queiroz Flanagan RJ, Taylor A, Watson ID, Whelpton R. 2007. Sample preparation.
ME, Lanças FM. 2006. Optimization of the SPME parameters and its In: flanagan RJ, Cuypers E, Maurer HH, Whelpton R, editors.
online coupling with HPLC for the analysis of tricyclic antidepressants Fundamentals of analytical toxicology. Chichester: Wiley; p. 49–94.
in plasma samples. J Chromatogr Sci. 44(6):340–346. doi: 10.1093/ Franco de Oliveira S, Zucoloto AD, de Oliveira CDR, Hernandez EMM,
chromsci/44.6.340. Fruchtengarten LVG, de Oliveira TF, Yonamine M. 2019. A fast and
Amitai Y, Erickson T, Kennedy EJ, Leikin JB, Hryhorczuk DO, Noble J, simple approach for the quantification of 40 illicit drugs, medicines,
Hanashiro PK, Frischer H. 1993. Tricyclic antidepressants in red cells and pesticides in blood and urine samples by UHPLC-MS/MS. J Mass
and plasma: correlation with impaired intraventricular conduction in Spectrom. 54(7):600–611. doi: 10.1002/jms.4369.
acute overdose. Clin Pharmacol Ther. 54(2):219–227. doi: 10.1038/clpt. Fu S, Lewis J, Wang H, Keegan J, Dawson M. 2010. A novel reductive
1993.133. transformation of oxazepam to nordiazepam observed during enzym­
Armbruster DA, Pry T. 2008. Limit of blank, limit of detection and limit atic hydrolysis. J Anal Toxicol. 34(5):243–251. doi: 10.1093/jat/34.5.243.
of quantitation. Clin. Biochem. Rev. 29(Suppl 1):S49–S52. Garland WA. 1977. Quantitative determination of amitriptyline and its
Bazregar M, Rajabi M, Yamini Y, Saffarzadeh Z, Asghari A. 2016. Tandem principal metabolite, nortriptyline, by GLC-chemical ionization mass
dispersive liquid-liquid microextraction as an efficient method for spectrometry. J Pharm Sci. 66(1):77–81. doi: 10.1002/jps.2600660119.
determination of basic drugs in complicated matrices. J Chromatogr Ito R, Ushiro M, Takahashi Y, Saito K, Ookubo T, Iwasaki Y, Nakazawa H.
A. 1429:13–21. doi: 10.1016/j.chroma.2015.11.087. 2011. Improvement and validation the method using dispersive
Borges EM, Goraieb K, Collins CH. 2012. O Desafio de analisar solutos liquid-liquid microextraction with in situ derivatization followed by
b�asicos por cromatografia l�ıquida em modo reverso: algumas gas chromatography-mass spectrometry for determination of tricyclic
TOXICOLOGY MECHANISMS AND METHODS 13

antidepressants in human urine samples. J Chromatogr B Analyt Saldanha GA, Pacheco ALB, Pego AMF, Bairros AV. 2022. Analysis of ben­
Technol Biomed Life Sci. 879(31):3714–3720. doi: 10.1016/j.jchromb. zodiazepines in plasma samples by DLLME and LC-DAD: critical
2011.10.012. aspects, flaws and issues encountered – a discussion. Quim Nova. 45:
Karpinska J, Starczewska B. 2002. Simultaneous LC determination of 1–8.
some antidepressants combined with neuroleptics. J Pharm Biomed Samanidou VF, Nika MK, Papadoyannis IN. 2007. Development of an
Anal. 29(3):519–525. doi: 10.1016/s0731-7085(02)00097-3. HPLC method for the monitoring of tricyclic antidepressants in bio­
Liebelt EL, Francis PD. 2002. Chapter 57 – Cyclic antidepressants. In: fluids. J Sep Sci. 30(15):2391–2400. doi: 10.1002/jssc.200700142.
Goldfrank LR, Flomenbaum NR, Lewin NA, Howland MA, Hoffman RS, Santos MG, Tavares IM, Barbosa AF, Bettini J, Figueiredo EC. 2017.
Nelson LS, editors. Goldfrank’s toxicological emergencies, 7th ed. USA: Analysis of tricyclic antidepressants in human plasma using online-
McGraw-Hill; p. 847–864. restricted access molecularly imprinted solid phase extraction fol­
Lundstrøm NH, Holgersen NK, Haastrup MB. 2022. The effect of smoking lowed by direct mass spectrometry identification/quantification.
on the plasma concentration of tricyclic antidepressants: a systematic Talanta. 163:8–16. doi: 10.1016/j.talanta.2016.10.047.
review. Acta Neuropsychiatr. 34(1):1–9. doi: 10.1017/neu.2021.28. Seidi S, Yamini Y, Rezazadeh M. 2013. Combination of electromembrane
Malfar�a WR, Bertucci C, Costa Queiroz ME, Dreossi Carvalho SA, Pires extraction with dispersive liquid-liquid microextraction followed by
Bianchi MdL, Cesarino EJ, Crippa JA, Costa Queiroz RH. 2007. Reliable gas chromatographic analysis as a fast and sensitive technique for
HPLC method for therapeutic drug monitoring of frequently pre­ determination of tricyclic antidepressants. J Chromatogr B Analyt
scribed tricyclic and nontricyclic antidepressants. J Pharm Biomed Technol Biomed Life Sci. 913-914:138–146. doi: 10.1016/j.jchromb.
Anal. 44(4):955–962. doi: 10.1016/j.jpba.2007.04.005. 2012.12.008.
Manousi N, Samanidou VF. 2020. Recent advances in the HPLC analysis Shamsipur M, Mirmohammadi M. 2014. High performance liquid chroma­
of tricyclic antidepressants in bio-samples. Mini Rev Med Chem. 20(1): tographic determination of ultra traces of two tricyclic antidepressant
24–38. doi: 10.2174/1389557519666190617150518. drugs imipramine and trimipramine in urine samples after their dis­
Martins ML, Primel EG, Caldas SS, Prestes OD, Adaime MB, Zanella R. persive liquid-liquid microextraction coupled with response surface
2012. Microextraç~ao L�ıquido-L�ıquido Dispersiva (DLLME): fundamentos optimization. J Pharm Biomed Anal. 100:271–278. doi: 10.1016/j.jpba.
e aplicaço~es. SC. 4(1):29–45. doi: 10.4322/sc.2012.004. 2014.08.008.
Meatherall R. 1994. Optimal enzymatic hydrolysis of urinary benzodi­ Silveira GO, Lourenço FR, Fonseca Pego AM, Guimar~aes Dos Santos R,
azepine conjugates. J Anal Toxicol. 18(7):382–384. doi: 10.1093/jat/18. Rossi GN, Hallak JEC, Yonamine M. 2021. Essential oil-based dispersive
7.382. liquid-liquid microextraction for the determination of N,N-dimethyl­
Mercolini L, Mandrioli R, Finizio G, Boncompagni G, Raggi MA. 2010. tryptamine and b-carbolines in human plasma: a novel solvent-free
Simultaneous HPLC determination of 14 tricyclic antidepressants and alternative. Talanta. 225:121976. doi: 10.1016/j.talanta.2020.121976.
metabolites in human plasma. J Sep Sci. 33(1):23–30. doi: 10.1002/ SINITOX (Sistema Nacional de Informaço ~es To �xico-Farmacolo �gicas). 2020.
jssc.200900493. Evolution of registered cases of human poisoning by toxic agent (in
Mohebbi A, Farajzadeh MA, Nemati M, Sarhangi N, A, Mogaddam MR. Portuguese). [accessed 2023 July 10] Available from: https://sinitox.
2019. Development of green sodium sulfate-induced solidification of icict.fiocruz.br/sites/sinitox.icict.fiocruz.br/files//Brasil3_1.pdf.
floating organic droplets-dispersive liquid phase microextraction Skopp G. 2004. Preanalytic aspects in postmortem toxicology. Forensic
method: application to extraction of four antidepressants. Biomed Sci Int. 142(2-3):75–100. doi: 10.1016/j.forsciint.2004.02.012.
Chromatogr. 33(11):e4642. SWGTOX (Scientific Working Group for Forensic Toxicology). 2013.
Mohebbi A, Farajzadeh MA, Yaripour S, Afshar Mogaddam MR. 2018. Scientific Working Group for Forensic Toxicology (SWGTOX) standard
Determination of tricyclic antidepressants in human urine samples by practices for method validation in forensic toxicology. J Anal Toxicol.
the three-step sample pretreatment followed by HPLC-UV analysis: an 37:452–474.
efficient analytical method for further pharmacokinetic and forensic Tabani H, Shokri A, Tizro S, Nojavan S, Varanusupakul P, Alexovi�c M.
studies. Excli J. 17:952–963. 2019. Evaluation of dispersive liquid-liquid microextraction by cou­
Mohebbi A, Yaripour S, Farajzadeh MA, Afshar Mogaddam MR. 2018b. pling with green-based agarose gel-electromembrane extraction: an
Combination of dispersive solid phase extraction and deep eutectic efficient method to the tandem extraction of basic drugs from bio­
solvent-based air-assisted liquid-liquid microextraction followed by logical fluids. Talanta. 199:329–335. doi: 10.1016/j.talanta.2019.02.078.
gas chromatography-mass spectrometry as an efficient analytical Theurillat R, Thormann W. 1998. Monitoring of tricyclic antidepressants in
method for the quantification of some tricyclic antidepressant drugs human serum and plasma by HPLC: characterization of a simple,
in biological fluids. J Chromatogr A. 1571:84–93. doi: 10.1016/j. laboratory developed method via external quality assessment. J Pharm
chroma.2018.08.022. Biomed Anal. 18(4-5):751–760. doi: 10.1016/s0731-7085(98)00263-5.
Montenarh D, Hopf M, Maurer HH, Schmidt P, Ewald AH. 2014. Detection Titier K, Castaing N, Le-D� eodic M, Le-Bars D, Moore N, Molimard M.
and quantification of benzodiazepines and Z-drugs in human whole 2007. Quantification of tricyclic antidepressants and monoamine oxi­
blood, plasma, and serum samples as part of a comprehensive multi- dase inhibitors by high-performance liquid chromatography-tandem
analyte LC-MS/MS approach. Anal Bioanal Chem. 406(3):803–818. doi: mass spectrometry in whole blood. J Anal Toxicol. 31(4):200–207. doi:
10.1007/s00216-013-7513-x. 10.1093/jat/31.4.200.
Nabil AA, Nouri N, Farajzadeh MA. 2015. Determination of three antide­ Tomai P, Gentili A, Curini R, Gottardo R, Franco Tagliaro, Fanali S. 2021.
pressants in urine using simultaneous derivatization and temperature- Dispersive liquid-liquid microextraction, an effective tool for the
assisted dispersive liquid-liquid microextraction followed by gas chro­ determination of synthetic cannabinoids in oral fluid by liquid chro­
matography-flame ionization detection. Biomed Chromatogr. 29(7): matography-tandem mass spectrometry. J Pharm Anal. 11(3):292–298.
1094–1102. doi: 10.1002/bmc.3396. doi: 10.1016/j.jpha.2020.11.004.
Peliç~ao FA, De Martinis BS, Pissinate JF. 2018. Cap�ıtulo 20 – Amostras Uddin MN, Samanidou VF, Papadoyannis IN. 2008. Development and val­
Biolo �gicas em An�alises Forenses: matrizes Usuais (Urina e Sangue). In: idation of an HPLC method for the determination of benzodiazepines
Dorta DJ, Yonamine M, Costa JL, Martinis BS, editors. Toxicologia and tricyclic antidepressants in biological fluids after sequential SPE. J
Forense. S~ao Paulo: Blucher; p. 381–392. Sep Sci. 31(13):2358–2370. doi: 10.1002/jssc.200800079.
Perry PJ, Zeilmann C, Arndt S. 1994. Tricyclic antidepressant concentra­ UNODC (United Nations Office on Drugs and Crime). 2009. Guidance for
tions in plasma: an estimate of their sensitivity and specificity as a de validation of analytical methodology and calibration of equipment
predictor of response. J Clin Psychopharmacol. 14(4):230–240. used for testing of illicit drugs in seized materials and biological
Queiroz RH, Lanchote VL, Bonato PS, de Carvalho D. 1995. Simultaneous specimens. [accessed 2020 December 20] Available from: https://
HPLC analysis of tricyclic antidepressants and metabolites in plasma www.unodc.org/documents/scientific/validation_E.pdf.
samples. Pharm Acta Helv. 70(2):181–186. doi: 10.1016/0031- Vaghar-Lahijani G, Saber-Tehrani M, Aberoomand-Azar P, Soleimani M.
6865(95)00019-6. 2018. Extraction and determination of two antidepressant drugs in
14 D. G. BERLATO ET AL.

human plasma by dispersive liquid–liquid microextraction–HPLC. J liquid microextraction followed by high-performance liquid chroma­
Anal Chem. 73(2):145–151. doi: 10.1134/S1061934818020144. tography. J Pharm Biomed Anal. 49(2):572–578. doi: 10.1016/j.jpba.
Vander Heyden Y, Nijhuis A, Smeyers-Verbeke J, Vandeginste BG, Massart 2008.11.036.
DL. 2001. Guidance for robustness/ruggedness tests in method valid­ Yazdi AS, Razavi N, Yazdinejad SR. 2008. Separation and determin­
ation. J Pharm Biomed Anal. 24(5-6):723–753. doi: 10.1016/s0731- ation of amitriptyline and nortriptyline by dispersive liquid-liquid
7085(00)00529-x. microextraction combined with gas chromatography flame ioniza­
Xiong C, Ruan J, Cai Y, Tang Y. 2009. Extraction and determination of tion detection. Talanta. 75(5):1293–1299. doi: 10.1016/j.talanta.
some psychotropic drugs in urine samples using dispersive liquid- 2008.01.039.

You might also like