1352
Lacorn & Weiss
: J
ournal of
AOAC I
nternational
Vol. 98, No. 5, 2015
J. Criteria for Acceptance of the Standard Curve
The shape of the standard curve is shown in the quality
assurance certificate enclosed in the test kit. Absorbances may
vary between different runs (e.g., due to different temperatures
or analysts). However, the shape of the standard curve should
be similar to the one given in the quality assurance certificate.
Minimum requirements are as follows:
(
1
) OD at 450 nm for standard 1 higher than 0.8.
(
2
) OD values for standards should continuously decrease
with higher concentrations, especially when comparing
standard 1 (0 ng/mL) and standard 2 (20 ng/mL).
(
3
) An OD value for standard 1 that is much higher than
the OD value stated in the certificate could be an indication of
errors during pipetting or incubation.
Results and Discussion
Collaborative Study Results
After finishing the analysis, each participant sent the data to
the Study Coordinator. These results are given in Table 1. After
statistical analysis of the data set, three problem laboratories
were identified. Further review found Laboratory F did not
run the calibrators in duplicate determinations as directed.
Laboratory E found no difference between calibration standards
S1 and S2, and as a consequence, a high OD difference
between standards S4 and S5 led to an unusual curve shape. An
interview with Laboratory E also revealed technical problems
during sample preparation. Laboratory K had a variation in the
calibration curve that was too high, and an interview revealed
the possibility of gluten contamination in the laboratory and
incorrect pipetting. As a result of these deviations, all data from
Laboratories E, F, and K were excluded from the statistical
evaluation.
For sample 5 (naturally contaminated syrup), all values were
calculated by cubic spline. Due to the fact that some OD values
were below the OD values of standard 2 (10 ng/mL prolamin;
corresponds to concentration of 10 mg/kg in the sample), these
values were extrapolated by the software. For the gluten-free
samples 1 and 4 the RIDA
®
SOFT Win software returned only
a result of <10 mg/kg, and extrapolation led to unrealistic
values. To be able to use the results of the analysis of the gluten-
free samples 1 and 4 in the performance statistics, estimates
of concentration values for these samples were required. For
this purpose, the calibration curves were constructed by using
a second-order polynomial model and used to recalculate the
results for samples 1 and 4 (7). This calibration provided an
estimate of concentrations for the gluten-free samples (Tables 1
and 2).
Statistical Analysis and Discussion
The remaining data of 13 laboratories are shown in Table 2
and were used to calculate the necessary statistics. Only
three outlying values were identified according to AOAC
INTERNATIONAL guidelines (12). These are indicated in
Table 2 by the superscripts “c” (for a Cochran outlier) and
“d” (for a double Grubbs’ outlier). The performance statistics
without outliers are shown in Table 2015.05.
From the measured overall mean concentrations of the
gluten-containing samples, recovery rates were calculated.
Table 2. Gluten concentrations determined by R5 competitive ELISA after eliminating laboratories E, F, and K
Gluten concentration, mg/kg
a
1
b
2
3
4
5
6
7
Repeat
Lab
1
2
1
2
1
2
1
2
1
2
1
2
1
2
A
2.13 5.80
23.6 20.5
111.6 93.9
4.47 7.73
7.6 8.62
46.7 47.2
153.0 170.0
B
1.46 2.66
40.8 13.8 151.4 127.4 2.98 2.13
10.6 5.1
38.8 53.0
163.6 122.8
C
5.30 10.6
34.2
c
82.2
c
192.2 107.6 6.12 1.90
12.8 12.6
47.2 67.4
181.4 143.4
D
0.74 1.77
23.8 28.6 175.2 97.6 –3.35 –3.41
9.8 11.0
33.0 60.2
106.4 107.6
G
6.06 4.30
32.4 32.0 216.2
d
208.2
d
3.29 –2.34 15.0 14.0
46.8 85.4
192.8 203.0
H
7.02 1.56
44.4 26.2 145.6 32.8
5.79 3.17
20.5 16.1
38.8 31.1
94.6 88.9
I
–0.65 –1.33 22.2 13.8 101.2 64.4 –0.89 –0.62
5.4 4.2
35.8 45.0
118.4 75.0
J
–1.50 1.14
21.2 20.0 121.8 128.8 –0.73 –1.63
7.4 8.0
45.6 58.3
132.9 139.2
L
1.69 –0.33 39.8 49.0 224.8
d
228.8
d
–1.83 3.39
13.2 11.6
64.0 67.2
171.6 244.6
M –0.66 4.13
19.9 19.3 129.4 133.6 –2.27 –0.62 10.0 8.6
36.1 39.6
161.7 120.4
N
0.04 0.76
34.2 18.4
97.0 108.6 1.84 4.41
10.8 9.2
43.4 44.6
117.6 154.4
O
1.57 0.41
19.1 16.5
110.7 136.6 –0.11 1.26
11.7 8.2
51.3 46.3
152.8 164.6
P
5.96 0.84
25.4 24.8 149.4 111.2
1.54 1.33
12.6 10.8
38.2 46.2 194.8 111.2
a
The calculation of the concentrations of the gluten-containing samples 2, 3, 5, 6, and 7 was done on the basis of a cubic spline function using the
RIDA
®
SOFT Win software; the statistics of the gluten-free samples 1 and 4 were calculated on the basis of a second-order polynomial function;
values for blinded samples are given as repeat 1 or repeat 2.
b
For samples 1–7
see
Table 1.
c
Means outlier according to the Cochran test.
d
Means outlier according to the double Grubbs’ test.