5 / 71
M
astovska
et al
.:
J
ournal of
AOAC I
nternational
V
ol
.
98, N
o
. 2, 2015
479
Collaborative Study
Purpose
The purpose of this study was to evaluate the method’s
intralaboratory and interlaboratory performance and submit the
results to AOAC INTERNATIONAL for adoption as an Official
Method for the determination of PAHs in seafood.
Study Design
This study evaluated the method performance for
determination of 19 selected PAHs, including alkyl homologs
relevant to an oil spill contamination (
see
Table 1), in three
seafood matrixes: shrimp, oysters, and mussels, with five
different levels of BaP ranging from 2 to 50 µg/kg. Each matrix
had a varying mixture of three different BaP levels (“low,”
“mid,” and “high”). The other studied PAHs were added
at varying levels from 2 to 250 µg/kg to mimic typical PAH
patterns (Table 2). The fortified analytes in the three matrixes
were analyzed as blind duplicates at each level of BaP and
corresponding other PAH levels. In addition, a blank with no
added PAHs for each matrix was analyzed singly. The AOAC
official method guidelines for collaborative study procedures (3)
were followed for the preparation of the study and data analysis.
Test Sample Preparation
Blank mussel and oyster samples were homogenized with
liquid nitrogen and tested in duplicate by an independent
laboratory for potential contamination with the target PAHs.
During homogenization, portions of the blank matrixes were
spiked with 1,7-dimethylphenanthrene (1,7-DMP) at 40 and
80 µg/kg in the case of mussel and oyster, respectively. These
were utilized as a homogenization check throughout the
course of the study. The collaborators determined 1,7-DMP
along with the other 18 analytes, which were spiked into 10 g
sample portions placed in polypropylene centrifuge tubes by
the study direction team. Five different spiking levels were
made at varying PAH concentrations (Table 2), resulting in
three different duplicate spiked samples/matrix in addition to
a blank. Participants were supplied with the test samples ready
for analysis labeled with unique identification numbers. All test
samples were shipped frozen on dry ice with a material receipt
document to be returned to the Study Directors. The test samples
had to be stored in a freezer set to maintain at least –20 ± 10°C.
Test samples were to be analyzed after completion of laboratory
qualification and practice sample analysis.
Blank shrimp matrix (peeled, without head and tail, and
uncooked) was homogenized without the use of liquid nitrogen
using a blender. After testing for potential contamination with
the target PAHs, 10 g blank sample portions were placed in
polypropylene centrifuge tubes, which were sent to study
participants together with spiking solutions labeled with unique
identification numbers. Using instructions provided by the
Study Directors, participants fortified the blank shrimp samples
themselves on the day of the analysis.
Three different spiking levels were used at varying PAH
concentrations (Table 2), resulting in three different duplicate
spiked samples in addition to a blank (seven samples altogether).
The blank shrimp samples were shipped frozen on dry ice with a
material receipt document to be returned to the Study Directors.
The test samples had to be stored in a freezer set to maintain at
least –20 ± 10°C. The spiking solutions were to be stored in a
refrigerator set to maintain 5
±
3°C. (
Note
: This modification
of the shrimp test sample preparation protocol (as compared to
mussel and oyster) was made (after consultations with the SPSC
PAH Working Group and the AOAC Methods Committee on
PAHs ) due to potential stability issues discovered during the
practice sample analysis and follow-up experimentswith fortified
shrimp samples stored at different conditions. 3-Methylchrysene
(3-MC) had to be replaced by 6-methylchrysene (6-MC) in the
spiking and calibration solutions for shrimp samples due to
the unavailability of a 3-MC reference standard at the time of
preparation and shipment of the new set of shrimp samples to
the study participants.)
Laboratory Qualification
During the laboratory qualification phase, the collaborators
conducted the following seven steps. These steps were
necessary because the Study Directors allowed the use of
various GC/MS instruments, GC columns, silica SPE cartridges,
and evaporation techniques and equipment. Therefore,
performance-based criteria were developed to help laboratories
optimize their GC/MS, SPE cleanup, and solvent evaporation
conditions; check and eliminate potential PAH contamination
in their reagent blanks; and become familiar with the method.
Laboratory qualification and practice sample results had to be
approved by the Study Directors before proceeding with the test
sample analysis. Sixteen laboratories entered the qualification
phase, but only 10 of them (listed in the
Acknowledgments
section) completed the qualification successfully and/or
continued in the study.
(
1
) The first step was a GC separation test where participants
analyzed a composite PAH solution by GC/MS/MS to obtain a
baseline separation of BaP and benzo[
e
]pyrene (concentration
ratio of 1:5); at least 50% valley separation of anthracene
and phenanthrene (concentration ratio 1:2.5, evaluated for
the anthracene peak); and at least 50% valley separation for
benzo[
b
]fluoranthene, benzo[
j
]fluoranthene, and benzo[
k
]
fluoranthene (concentration ratio of 1:1:1).
(
2
) The second step was a calibration range test where
participants prepared calibration standards and obtained
normalized calibration curves for the studied PAHs versus
respective labeled internal standards (
13
C-PAHs). Collaborators
had to determine the linear range, test for carryover by injecting
a solvent blank after the highest standard, and adjust injection
conditions (such as injection volume, number of washes, syringe
size, etc.) to achieve low detection limits, acceptable linearity
for the tested concentration range, and minimum carryover.
Coefficient of determination (r
2
) values should be 0.990 or
greater, and back-calculated concentrations of the calibration
standards should not exceed ±20% of theoretical. For lower
concentration levels, a limited calibration curve (without
the higher-end concentration points) may be used for better
accuracy. If a well characterized quadratic relationship occurs,
then a best-fitted quadratic curve may be used for calibration.
Otherwise, if the back-calculated concentrations exceed ±20%
of theoretical, normalized signals of the nearest two calibration