112 / 363
1384
Bidlack et al.:
J
ournal of
AOAC I
nternational
V
ol.
98, N
o.
5, 2015
constructed from these standards, and the regression parameters
from least-squares fittings were used to back calculate the
concentration of each working standard to determine calibration
errors at each level. It should be noted that all commercially
available vitamin K
1
standards contain a mixture of
cis
and
trans
vitamin K
1
. The percentage of
trans
vitamin K
1
in the standard
was determined experimentally for each run using
cis
and
trans
peak areas from all working standard chromatograms. The
experimentally determined ratio of
trans
vitamin K
1
was then
used to calculate the
trans
vitamin K
1
standard concentrations
of the working standards.
Trans
vitamin K
1
LOD and LOQ were determined
experimentally by injecting a very low level vitamin K
1
standard of known concentration and measuring the S/N.
Trans
vitamin K
1
LOD and LOQ in the standard solution were
calculated by multiplying the background noise by 3 (LOD) or
10 (LOQ) and dividing by the sensitivity, which was defined
as the ratio of the analytical signal to the concentration of the
analyte producing the signal. Product LOD and LOQ were
extrapolated from the standard LOD and LOQ using a typical
sample weight and dilution volume.
Ruggedness (or robustness) was not explicitly studied;
however, several parameters relevant to this were varied during
the SLV in order to factor as much uncertainty as possible into
the method performance metrics. Samples were prepared by two
analysts and analyzed with silica columns from three different
vendors. New mobile phase and postcolumn reagents were
made daily, and two sets of stock, intermediate, and working
standards were prepared and used during validation.
All of the unfortified matrixes were expected to contain
some
trans
vitamin K
1
and could not be used to unambiguously
establish method specificity; however, this method uses a
very specific detection technique. Relatively few compounds
Table 3. Trans vitamin K
1
SLV data–accuracy
Spike level
100%
50%
Sample type
No. of replicates
(duplicates on multiple days)
Native level,
µg/100 g RTF Recovery, % RSD, % Recovery, % RSD, %
Child formula powder
6
2.66
98.2
5.9
96.2
7.1
Infant elemental powder
6
7.57
93.2
7.6
94.0
2.6
SRM 1849a
6
1.11
a
104
2.9
95.5
1.8
Adult nutritional powder milk protein based
6
3.26
96.4
2.0
95.1
1.8
Infant formula powder partially hydrolyzed milk based
6
7.69
96.6
3.6
91.9
3.0
Infant formula powder partially hydrolyzed soy based
6
8.99
97.9
1.0
96.1
2.6
Adult nutritional powder low fat
6
2.92
98.0
2.5
95.2
2.6
Infant formula powder milk based
6
6.09
97.6
1.0
102
2.3
Infant formula powder soy based
6
6.26
97.9
1.7
102
0.3
Infant formula RTF milk based
6
9.01
100
1.3
104
0.4
Adult nutritional RTF high protein
6
9.10
96.7
2.7
106
0.7
Adult nutritional RTF high fat
6
10.7
98.2
1.2
93.8
4.0
a
Results reported as mg/kg powder.
Table 4. Summary of
trans
vitamin K
1
relative (%) calibration errors by level (30 curves)
a
Calibration level
b
Mean
Median
Minimum
Maximum
P
c
1
0.247
1.55
–7.94
5.04
0.698
2
0.484
0.879
–3.65
2.93
0.0930
3
0.117
0.0280
–1.00
2.21
0.350
4
0.0972
–0.0710
–1.08
1.9
0.443
5
–0.328
–0.271
–1.70
1.47
0.0460
6
0.0561
0.101
–0.421
0.372
0.201
Run average
0.104
0.364
–1.73
1.18
0.428
a
r
2
for the 30 curves ranged from 0.99985 to 1.00000, with an average of 0.99994.
b
Levels 1–6 corresponds to
trans
vitamin K
1
concentrations of 2–3, 6–8, 11–13, 22–30, 37–45, and 74–88 µg/L.
c
P
value for one sample
t
-test relative to zero.
Candidates for 2016 Method of the Year
111