126 / 258
1400
Jing
et al.
: J
ournal of
AOAC I
nternational
Vol. 98, No. 5, 2015
(
d
) For this analysis (for either choline or carnitine), given
the mass of IS added to the samples and the concentration of IS
in the standards, the DF can be shown to be:
DF = 3.125/sample weight (g)
(2)
(
e
) Thus, the final result (C
x
) can be calculated by combining
Equations 1 and 2.
Validation Protocol
An SLV protocol was carried out to ensure that the
method meets the requirements of the SMPRs for choline
(SMPR
2012.013
) and carnitine (SMPR
2012.010
). The SMPRs
are summarized in Table 1. The full suite of 11 “SPIFAN
matrixes” were used in this SLV. These materials (Table 2
)
were
made by the manufacturers specifically for the SPIFAN project
to be representative of nearly the entire body of infant and
adult nutritional products. By SPIFAN convention, the powders
were all prepared as reconstitutions (11.1% by weight), and
the indicated aliquot sizes were taken. The three liquid/ready-
to-feed (RTF) samples were weighed as-is. The SLV probed
the usual parameters of linearity, LOQ/LOD, repeatability,
intermediate precision, spike recovery (for accuracy), and
agreement of results to independent methods or SRMs.
(
a
)
Linearity
.—On each of 3 different days, calibration
curves were prepared for choline and carnitine, and then
five independently made standard solutions of differing
concentrations were injected as samples. The standard
solutions covered the range of the calibration curve from 30 to
300 µg/L choline and from 8 to 80 µg/L carnitine, except the
carnitine linearity test included a solution at ½ WS1 and the
choline linearity test included a solution that was well above
WS3 in concentration. The 3-day mean concentration results
from these standard solutions were compared to their nominal
concentrations, with an expected agreement of better than ±5%.
(
b
)
LOD/LOQ.
—The adult powder was selected as a suitable
placebo in lieu of other possible placebos included in the
SPIFAN materials kit. The placebo was weighed at 3.00 g of a
10.93% (w/w) reconstitution and spiked with a known amount of
choline (100 µL of a 593.4 µg/mL solution = 59.3 µg, as choline
hydroxide in this case) and carnitine (61 µL of a 77.9 µg/mL
solution = 4.75 µg). These spike amounts are at the SMPR
required LOQ limit: approximately 60 µg/3.0 g RTF = 20 µg/g
= 2.0 mg choline hydroxide/100 g RTF; 4.75 µg/3.0 g RTF =
0.16 mg/100 g RTF. The placebo was analyzed seven times
to establish a baseline value, and then seven spiked placebos
were analyzed and compared to the baseline to determine spike
recovery. A spike recovery of 90–110% would establish the
spike level as at, or above, the required LOQ.
(
c
)
Accuracy (trueness).—
SRM 1849a was analyzed with
this method, and results were compared to certified values for
both choline and carnitine. The 11 SPIFAN matrixes were also
spiked with known amounts of choline and carnitine at two
different levels, 40 and 80% of the baseline product level, to
ensure spike recoveries were in the 90–110% range as required
by the SMPR. The spiking was performed on 3 different days,
and duplicate spike samples were prepared on each day, at each
spiking level. For carnitine, the whole experiment was repeated
with acetylcarnitine (Carbosynth LLC, >98% purity, San Diego,
CA) instead of L-carnitine, to prove reduction of that species to
carnitine with the method’s basic hydrolysis. Finally, accuracy
of this method was established by comparison of choline results
to independent methods. AOAC
999.14
results were obtained
from a commercial laboratory (Covance Laboratories, Madison,
WI) over 2 days, in duplicate. AOAC
2012.20
results were
obtained from the Thermo/Dionex website (5).
(
d
)
Precision
.—Each of the 11 SPIFAN products was
analyzed by the method on 6 days in duplicate, for both choline
and carnitine, running the method for both free analyte (no
microwave digestion or subsequent basic hydrolysis) and total
choline/carnitine. The data were collected by two analysts on
two different instruments (Waters Acquity TQD, and Xevo
TQS), splitting the days evenly among the analyst/instruments.
The repeatability precision and the intermediate precision were
calculated from the resulting data using analysis of variance
calculations from Microsoft (Redmond, WA) Excel.
(
e
)
Specificity
.—Secondary ion transitions were monitored
throughout the validation and assessed afterwards for the
specificity information they may provide.
Validation Results
Chromatography
.—Example chromatograms for the SRM
1849a analysis are shown in Figure 1. Choline and its IS elute
first at about 2.5 min followed by the carnitine and its IS at
about 4.4 min. Four additional traces are shown in Figure 1
representing the secondary ion transitions that were followed
for specificity verification. These data are discussed at the end
of this report.
Linearity
.—The lowest correlation coefficient observed
over the course of the study was 0.9992 for either choline
or carnitine. Table
3
shows the results of the 3-day linearity
study. The average recovery of independent standard solutions
run as samples was 99.1–101.2% at various points along the
calibration curve, meeting SMPRs. Note that the method should
not be used outside this proven region of linearity. Thus, the
lowest concentrations checked (3.4 µg/L carnitine and 30 µg/L
choline) become the practical LOQ for the method, even if the
Table 3. Linearity results
a
Nominal carnitine, µg/L
Recovery, % (avg. 3 days)
RSD, %
3.4
101.2
1.8
6.9
99.3
1.2
17.2
100.3
1.1
34.5
100.0
1.5
51.7
99.4
0.7
68.9
100.1
1.5
Nominal choline, µg/L
Recovery, % (avg. 3 days)
RSD, %
29.7
100.3
2.0
59.3
99.6
1.7
119
100.3
2.2
237
99.3
1.4
356
100.1
2.4
510
99.4
1.1
a
Calibration curve was 7.5–75 µg/L carnitine and 30–300 µg/L choline.
126