Validation of a fast screening method for the detection [601511]

ORIGINAL PAPER
Validation of a fast screening method for the detection
of cocaine in hair by MALDI-MS
Susanna Vogliardi &Donata Favretto &
Giampietro Frison &Sergio Maietti &Guido Viel &
Roberta Seraglia &Pietro Traldi &Santo Davide Ferrara
Received: 24 September 2009 /Revised: 30 November 2009 /Accepted: 7 December 2009 /Published online: 10 January 2010
#Springer-Verlag 2010
Abstract The sensitivity and specificity of a novel method
of screening for cocaine in hair, based on matrix-assistedlaser desorption/ionisation (MALDI) mass spectrometry
(MS), have been evaluated. The method entails a rapid
extraction procedure consisting of shaking 2.5 mg pulv-erised hair at high frequency in the presence of an acidic
solution (160 µL of water, 20 µL of acetonitrile and 20 µL
of 1 M trifluoroacetic acid) and a stainless-steel bullet.Following centrifugation, the supernatant is dried under a
nitrogen stream, and the residue is reconstituted in 10 µL
of methanol/trifluoroacetic acid (7:3; v/v). One microlitre
of the extract is deposed on a MALDI sample holder
previously scrubbed with graphite; an α-cyano-4-hydrox-
ycinnamic acid (matrix) solution is electrosprayed over thedried sample surface to achieve a uniform distribution of
matrix crystals. The identification of cocaine is obtained by
post-source decay experiments performed on its MH
+ion
(m/z304), with a limit of detection of 0.1 ng/mg of cocaine.
A total of 304 hair samples were analysed in parallel by
MALDI-MS and a reference gas chromatography-MSmethod. The obtained results demonstrate specificity and
sensitivity of 100% for MALDI-MS. Evidence of cocainepresence was easily obtained even when hair samples
exhibiting particularly low cocaine levels (<0.5 ng/mg)
were analysed.
Keywords Hair.Cocaine .MALDI .Screening test .
Validation .Drug screening .Hair analysis .
MALDI-MS
Introduction
Hair testing is routinely used as a powerful tool for the
detection of drug traces in many clinical and forensic
toxicology applications. Hair is a special matrix for the
retrospective investigation of chronic drug abuse, capable ofdemonstrating even a single administration [ 1]. Additionally,
segmental hair analysis can yield precious information about
the time course of the substance consumption [ 2].
Considering the increasing abuse of cocaine (COC),
currently one of the most widespread drugs of abuse in the
world, it has become important to adopt simple and fastscreening methods to detect COC in a large number of hair
samples. These methods are not necessarily expected to
provide quantitative data; they should however exhibitdetection limits sufficient to allow an initial discrimination
between negative and positive samples that should then be
confirmed and quantified by classical chromatographic-
mass spectrometric methods [ 3]. Matrix-assisted laser
desorption/ionisation-mass spectrometry (MALDI-MS),due to inherent characteristics such as the soft ionisation,
the privileged production of molecular species, the better
tolerance with respect to electrospray ionisation-MS tointerference from salts and buffers and the simplicity ofElectronic supplementary material The online version of this article
(doi:10.1007/s00216-009-3387-3 ) contains supplementary material,
which is available to authorized users.
S. Vogliardi :D. Favretto :G. Frison :S. Maietti :G. Viel :
S. D. Ferrara ( *)
Forensic Toxicology and Antidoping,
University-Hospital of Padova,
Via Falloppio 50,35121 Padova, Italy
e-mail: santodavide.ferrara@unipd.it
R. Seraglia:P. Traldi
CNR, ISTM,
Corso Stati Uniti 4,
35100 Padova, ItalyAnal Bioanal Chem (2010) 396:2435 –2440
DOI 10.1007/s00216-009-3387-3

sample preparation, is ideally suited for simultaneous, rapid
and high-throughput analyses of complex mixtures and
might be employed as a screening method for drugdetection in hair.
In a previous paper [ 4], MALDI-MS and post-source
decay (PSD) experiments were employed to develop afast screening method for the detection of COC and its
metabolites benzoylecgonine (BE) and cocaethylene
(CE) in hair. That approach was utilised to analysetwo hair samples whose content, as determined by gas
chromatography-MS (GC-MS), was 9.5 and 100 ng/mg
of COC, respectively. The obtained results demonstratedthe feasibility of using the MALDI technique for
detecting drugs and their metabolites in a complex
biological matrix. However, the MALDI-MS methodrequired to be tested on a large sample number before
being proposed as a fast screening test to detect COC in
hair samples, particularly in specimens exhibiting lowCOC concentrations. There fore, after optimising the
sample preparation procedure and estimating the limit
of detection of the method, 304 hair samples wereindependently analysed by a GC-MS reference method
[4] and by MALDI-MS, using GC-MS as the gold
standard. The obtained results (251 true negative and 53
true positive hair samples) were used to evaluate the
performances of the MALDI-MS method.
Experimental
Materials
Analytical-grade acetone, methanol, n-hexane, sodium
dodecyl sulphate (SDS) and trifluoroacetic acid (TFA) were
obtained from Merck (Darmstadt, Germany). Water wasprovided by a Milli-Q Plus system (Millipore, Molsheim,
France).
α-Cyano-4-hydroxycinnamic acid (HCCA) was obtained
from LBL (LaserBioLabs, Sophia-Antipolis, France).
Hair samplesDrug-free hair samples were collected from the laboratory
personnel. Head and pubic hair samples (304 samples) usedin this study were collected for clinical and forensic toxicol-
ogy purposes at the Forensic Toxicology and Antidoping Unit
of the University Hospital of Padova.
Certified Standard Reference Material (NIST-SRM-2379),
consisting of human hair containing COC (7.45 ng/mg),
BE (4.01 ng/mg), CE (2.67 ng/mg), methamphetamine(5.20 ng/mg) and phencyclidine (6.24 ng/mg), was
obtained from the National Institute of Standard Tech-
nology (NIST, Gaithersburg, MD, USA).Hair sample pre-treatment
Hair samples were decontaminated by washing whole
strands in 10 mL of a 10% ( w/v) aqueous solution of
SDS, followed by washes in 10 mL of water (twice) and
10 mL of acetone. After washing, samples were pulverisedin a MM400 mixer mill (Retsch, Haan, Germany) at the
amplitude of 30 Hz for 5 min.
MALDI
Hair sample preparation for MALDI-MS analysis
Of pulverised hair samples, 2.5 mg aliquots were accurately
weighted in small polypropylene vials (1 mL volume).
After the addition of one stainless-steel bullet, 160 µL of
water, 20 µL of acetonitrile and 20 µL of TFA 1 M, thevials were capped and placed into the jar of the mixer mill
that was operated at the amplitude of 30 Hz for 5 min.
Following centrifugation, the supernatant was transferred toa 2-mL crimp vial and dried under nitrogen stream. The
residue was reconstituted in 10 µL of methanol/TFA (7:3; v/
v) solution. One-microlitre aliquots of the obtained sol-
utions were deposed on the MALDI stainless-steel sample
holder treated with graphite from pencil B and let to dry on
air, then the HCCA matrix solution was sprayed over, by
means of the sieve-based device (SBD), employing a 38-
µm, 450 mesh sieve and a flow rate of 5 µL/min for 1 min.
MALDI-MS analyses
MALDI measurements were performed using a Bruker
Ultraflex II instrument (Bruker Daltonics, Bremen, Ger-
many), operating in reflectron positive ion mode. Thepulsed ion extraction conditions were as follows: IS1,
25 kV; IS2, 21.70 kV; reflectron, 26.30 kV; delay time,
150 ns. The signal of 100 shots was accumulated to obtaina spectrum. External mass calibration was performed on the
basis of [M+H]
+and [2M+H]+ions of HCCA at m/z
190.0498 and 379.0924, respectively. The ion selection andmass calibration for the PSD experiments were performed
using the FAST
®method.
Validation protocol for MALDI-MS
The evaluated performance parameters of the proposed
qualitative method were confirmation of identity, selectiv-
ity, lower limit of detection (LLOD) and precision (false
positive rate, false negative rate, sensitivity and specificity).
Confirmation of identity was studied by PSD experi-
ments, whereas the selectivity was studied by analysing
blank (for COC) hair, namely, drug-free hair collected from2436 S. Vogliardi et al.

the laboratory personnel and hair samples containing drugs
other than COC.
The LLOD was determined by repeated analysis of
certified reference material (NIST-SRM-2379) containing
COC at 7.45 ng/mg diluted with blank hair in order to
obtain progressively lower COC concentrations; ten repeat-ed measurements were performed at each concentration.
Sensitivity, specificity, false positive rate and false
negative rate, to be used to assess the precision of themethod, were obtained by comparison with the GC-MS
quantitative method and were defined as follows:
False positive rate ¼
fp
tnțfp;
False negative rate ¼fn
tpțfn;
Sensitivity ¼tp
tpțfn;
Specificity ¼tn
tnțfp:
where tn=true negative samples, fn=false negative, tp=true
positive and fp=false positive.
Results and discussion
Optimization of the MALDI-MS procedure
The qualitative MALDI-MS method studied in our previous
investigation was able to evidence COC and its metabolites
BE and CE in hair in the range 10 –100 ng/mg, with as low
as 1 mg of sample [ 4]. In that occasion, however, we
focused our attention on optimising the MALDI-MS
procedure for the detection of COC, BE and CE and did
not evaluate the method in terms of sensitivity andselectivity.
Nonetheless, following guid elines for forensic hair
analysis [ 5], if a screening test is used, an appropriate
method validation including calibrators and controls in a
hair matrix must be performed, and analytes of interest
must be identified to minimise false negatives; the samerecommendations suggest 0.5 ng/mg COC as a “cut off ”for
a screening test, i.e. a COC level of 0.5 ng/mg must
produce a positive result. It was therefore mandatory toverify the performance of the MALDI-MS method before
proposing it as a screening test.To reach a limit of detection ≤0.5 ng/mg, the previously
described extraction procedure was considered inappropri-
ate, and a new method, developed in our laboratory formulti-analyte hair analysis by liquid chromatography-high-
resolution MS, was applied. The new procedure, particu-
larly suitable when tiny amounts of hair are available,consists of placing 2.5 mg of pulverised hair samples in a
polypropylene vial containing 200 µL of an acidic solution
and one hair stainless-steel bullet; the vial is placed in thejar of an automatic mixer/pulveriser oscillating at 30 Hz for
5 min [ 6]. In these conditions, both the pulverisation and
extraction steps take place at the same very short time. Theextraction probably exploits the energy provided by
vibration and mechanical pulverisation against the walls
of the plastic vial and the surface of the bullet. Onlycentrifugation is needed to obtain a clear aqueous superna-
tant. In order to improve the sensitivity of the MALDI-MS
method and achieve the detection of COC levels lower than1 ng/mg in 2.5 mg hair, the concentration (20:1) of the
extracts by drying the aqueous supernatants and reconsti-
tuting the residues in 10 µL of methanol/TFA (7:3; v/v) was
considered necessary. One-microlitre aliquots of reconsti-
tuted extracts were deposed on the MALDI stainless-steel
sample holder scrubbed with graphite from pencil B [ 7].
Once the solutions have dried, the HCCA matrix solution is
electrosprayed over them at a flow rate of 5 µL/min for1 min, by means of the SBD [ 8]. In these conditions, a
uniform and discrete distribution of matrix crystals on the
sample surface is realised [ 4].
This procedure was applied to the detection of COC in
calibrators prepared by diluting certified reference material
(NIST-SRM-2379) with drug-free hair and in positive andnegative samples (as determined by a reference GC-MS
method [ 4]). The same GC-MS procedure for the detection
of COC and metabolites was applied to NIST-SRM-2379,obtaining the following results: 7.40 ng/mg COC, 3.90 ng/
mg BE and 2.61 ng/mg CE.
In Fig. 1, the MALDI spectra of a negative sample (A)
and two positive samples (B and C) whose COC content, as
determined by the GC-MS standard quantitative procedure,
is 47.0 and 4.8 ng/mg, respectively, are shown. The ionicspecies at m/z304 (MH
+of COC) are detected in both B
and C, whereas it is not measurable at all in the negative
sample; however, sample C, whose COC content inferredby GC-MS is an order of magnitude lower than sample B,
gives rise to a MALDI spectrum of poorer quality, with a
low signal-to-noise (s/n) ratio.
It is known, from the early days of collisional spectroscopy,
that MS/MS experiments lead to a concomitant increase of
specificity, being able to put in evidence characteristicfragments of the analyte with a significant suppression of the
chemical background [ 9]. In MALDI instruments, MS/MS
data can be obtained by PSD experiments [ 10], and thisValidation of fast screening for detection of cocaine in hair 2437

approach was in fact employed to obtain the product ion
spectra of the ionic species at m/z304.
In Fig. 2, the PSD spectra of the ion at m/z304 for
samples A, B and C are reported. It is at first sight evident
that only for B and C an intense peak at m/z182 is
detectable with an s/n ratio in the order of 100. That
product ion is a characteristic fragment of COC structure,
being detected either in collisional regimes [ 11]o ri nE I
conditions [ 12] and can be justified by the fragmentation
pathway reported in Fig. 3. In MALDI conditions, the ion
atm/z182 originates from MH+through the loss of benzoic
acid, in agreement with the even electron rule [ 13], while in
the case of M+species generated in EI conditions, its
formation is due to radical loss of C 6H5CO 2. In the case of
the negative sample A, this fragment ion is not detectable,
as proved by the low ion intensity and the high electrical
noise present in the same time window.
As already discussed in a previous paper [ 14], PSD is
different from a “classical ”MS/MS experiment entailing
ion selection, collision with a target gas in a well-definedspatial region and analysis of the collisionally generated
fragment ions. However, the detection of product ions can be
explained by phenomena typical of the MALDI experiment.The PSD method just defines a time window allowing the
selection of the precursor ions whose internal energy
distribution is quite wide, due to collision phenomena
occurring in the cloud generated by the laser irradiation [ 10].
The performance of the method was subsequently chal-
lenged with the analysis of eight samples in the range 0.3 –
0.5 ng/mg (results reported as “Electronic supplementary
material ”). In most cases, the MH
+ion of COC at m/z304exhibits a low s/n ratio; however, when PSD experiments are
performed, the ionic species at m/z182 is always detected
with s/ n>3. It is particularly important to be able to detect
COC in hair at or below the level of 0.5 ng/mg, since that
concentration has been proposed as a screening “cut off ”by
several guidelines, i.e. a COC level of 0.5 ng/mg must
produce a positive result [ 3,5].
As to BE and CE, their MALDI-MS detection posed
problems reasonably due to their lower and variable (8 –
40%) abundance in hair with respect to the parent
compound COC. Since the developed MALDI-MS methodwould be used as a screening test prior to GC-MS or liquid
chromatography-MS confirmatory analysis and the parent
compound is always the most abundant component in COCpositive hair [ 3], the identification of metabolites was not
further investigated.
Validation
For the validation of the proposed qualitative method, i.e. a
method of analysis whose response is either the presence or
absence of the analyte, detected either directly or indirectly
in a certain amount of sample [ 15], the guidelines of
EURACHEM [ 16] and Association of Analytical Commu-
nities [ 15] were followed.
The confirmation of identity was reached by the above-
described PSD results; consequently, COC was positively
identified in MALDI experiments when a PSD spectrum ofions at m/z304 gave rise to a product ion at m/z182 with an
s/n ratio ≥3 .D u et ot h ei n t r i n s i cn a t u r eo ft h eM A L D I
experiments, spectra can be acquired more than once on
Fig. 1 Matrix-assisted laser
desorption/ionisation spectra
obtained for negative sample A
(a), positive sample B at
47.0 ng/mg ( b) and positive
sample C at 4.8 ng/mg ( c)2438 S. Vogliardi et al.

the same sample; variability of the s/n ratio was observed
in the range 5 –20% for positive samples containing COC
in the range 0.1 –50 ng/mg when ten different spectra were
acquired for the same sample, accumulating the signalsoriginated by 100 laser shots for each spectrum.
The analysis of 30 hairs from different sources (includ-
ing drug-free hair and hair from people exposed to drugsother than COC) was used as an evaluation for selectivity,i.e. the ability to measure an analyte in the presence of
endogenous compounds, hair matrix components and other
drugs. No interference was evidenced.
The LLOD, defined as the lowest concentration of the
analyte which the test can reliably detect as positive in the
given matrix using the above reported criterion for COC
identification, turned out to be 0.1 ng/mg by repeatedanalysis (ten independent replicates) of reference material
(NIST-SRM-2379) diluted with blank hair to achieve
progressively lower COC concentrations: 7.40, 3.70, 1.85,
0.74, 0.40, 0.20, 0.10, 0.07 and 0.05 ng/mg, respectively. In
Fig. 4, the response curve obtained by plotting concentra-
tion versus percentage of positive results is reported; the
threshold concentration at which the test becomes reliable is
0.10 ng/mg.
Subsequently, 304 head and pubic hair samples that were
collected at our laboratory for forensic toxicology purposes
Fig. 3 Fragmentation pathway from the ionic species at m/z304 to
m/z182
Fig. 2 Product ion spectra
obtained by post-source decay
experiments conducted on the
ion at m/z304 for negative
sample A ( a) and positive sam-
ples B ( b) and C ( c)
Fig. 4 Response curve of percentage of positive results versus analyte
concentration. Ten independent measurements were performed at each
concentrationValidation of fast screening for detection of cocaine in hair 2439

were independently analysed by the MALDI-MS and the
reference GC-MS method for the purposes of building a
contingency table. The employed GC-MS procedure, whichis routinely used in our lab, exhibits a lower limit of
quantification of 0.1 ng/mg and a limit of detection of
0.05 ng/mg; we decided to declare “positive ”a sample with
a COC level ≥0.1 ng/mg. Through this criterion, 251
samples were found to be negative, and 53 samples resulted
positive. The distribution of COC concentrations amongpositive head and pubic ha ir samples is attached as
“Electronic supplementary material ”.
The results obtained by the two methods were compared
and used to calculate true positives, true negatives, false
positives and false negatives. Neither false positive nor
false negative samples were detected; the sensitivity,defined as the percentage of true positive (by GC-MS)
specimens identified as positive by MALDI-MS, is 100%,
and the specificity, defined as the percentage of specimensthat were not true positives (by GC-MS) and were
identified as negative by MALDI-MS, is 100%. The false
positive rate and false negative rate are 0%, and theywitness the precision of the method [ 17,18].
As to the positive samples, the excellent performance of
the MALDI-MS method at low analyte levels, with as low
as 2.5 mg hair, was proven by the positive identification of
samples with COC levels in the range 0.3 –0.5 ng/mg.
In conclusion, the proposed analytical procedure
proved to be fit for purpose, i.e. it is valuable for a fast
screening of COC in hair. When compared to other screen-ing methods, such as immunochemical assays, its strong
points are
–Only 2.5 mg hair and 5 min extraction are needed prior
to MALDI-MS measurements.
–The MALDI-MS analysis, with its PSD features, assures
molecular recognition of COC, i.e. the target analyte.–The characteristics of MALDI sample holders allow
high-throughput routines (about 100 samples/holders,
depending on the specific instrument).
–Sensitivity and specificity of 100% and false positive
and false negative rates of 0% are obtained when 304
samples were analysed by both the MALDI-MS andthe reference GC-MS method.
References
1. Musshoff F, Madea B (2007) Anal Bioanal Chem 388:1475 –1494
2. Kintz P, Villain M, Cirimele V (2006) Ther Drug Monit 28:442 –446
3. Pragst F, Balikova MA (2006) Clin Chim Acta 370:17 –49
4. Vogliardi S, Favretto D, Frison G, Ferrara SD, Seraglia R, Traldi P
(2009) J Mass Spectrom 44:18 –24
5. Society of Hair Testing (2004) Forensic Sci Int 145:83 –84
6. Favretto D, Stocchero G, V ogliardi S, Frison G, Trevisanuto D,
Castagna F, Ferrara SD (2009) Neonatal hair analysis by liquid
chromatography-high-resolution mass spectrometry to reveal gesta-tional exposure to venlafaxine. Ther Drug Monit. doi: 10.1097/
FTD.0b013e3181c01e16
7. Langley GJ, Herniman JM, Townell MS (2007) Rapid Commun
Mass Spectrom 21:180 –190
8. Molin L, Cristoni S, Crotti S, Bernardi LR, Seraglia R, Traldi P
(2008) J Mass Spectrom 43:1512 –1520
9. Grant ER, Cooks RG (1990) Science 250:61 –68
10. Kaufmann R, Spengler B, Lu tzenkirchen F (1993) Rapid
Commun Mass Spectrom 7:902 –910
11. Wang PP, Bartlett MG (1998) J Mass Spectrom 33:961 –967
12. Kintz P, Mangin P (1995) Forensic Sci Int 73:93 –100
13. Karni M, Mandelbaum A (1979) Organic Mass Spectrom 15:53 –64
14. Moneti G, Francese S, Mastrobuoni G, Pieraccini G, Seraglia R,
Valitutti G, Traldi P (2007) J Mass Spectrom 42:117 –126
15. Feldsine P, Abeyta C, Andrews WH (2002) J AOAC Int 85:1187 –
1200
16. EURACHEM (1998) A laboratory guide to method validation and
related topics, Teddington, Middlesex, UK. http://www.eurachem.
ul.pt. Accessed 21 Sept 2009
17. Ellison SLR, Fearn T (2005) Trends Anal Chem 24:468 –476
18. Trullols E, Ruisánchez I, Rius FX (2004) Trends Anal Chem
23:137 –1452440 S. Vogliardi et al.