The feasibility of harmonizing gluten ELISA measurements

Figures

• 

  Fig. 1. Youden plots showing the z scores derived from robust averages and standard deviation set at 25% of the reference value. The pairs of samples are matched for similar matrices. The z scores obtained for one of the samples are plotted against the z scores obtained for the other sample. Ellipses represent the mean ± 2 standard deviations for the results of each kit. A: 1st study data; B: 2nd study data; C: 1st study data corrected with universal calibrator, D: 2nd study data corrected with BME-IGRM. Two samples with largest variability in results resulting in extreme z scores (meat sample and cereals sample) were excluded from the graphs presenting 2nd study data.

• 

  Fig. 2. Assessment of the qualitative agreement between results before and after correction with BME-IGRM. Column plots show number of statements with regards to the gluten content (gluten-free, low gluten, contains gluten) per sample before (A) and after (B) correction with a BME-IGRM matrix material. Scatter plots show the response curves representing rate of “ > 20 mg kg−1” or “ > 100 mg kg−1” responses before (C, D) and after (E, F) correction with BME-IGRM.

• 

  Fig. 3. Comparison of efficiency of extraction with the kit-supplied and generic procedure. Relative differences between gluten concentration measured following the kit buffer extraction (Ckit) and gluten concentration measured following the generic buffer extraction (Cgen). Error bars represent the expanded uncertainty. The results were expressed as the averages of the relative differences between gluten content measured following the kit buffer extraction and gluten content measured following the generic buffer extraction.

• 

  Fig. 4. Examples of results for assessing the commutability of the BME-IGRM material for: A) where the material is commutable; B) where the material is non-commutable; and C) where the assessment is ambiguous due to a large spread of results around the regression line. Results for food samples are shown as black squares, results for BME-IGRM are shown as empty circles. Solid lines represent the regression line plotted for the food samples, dashed lines represent the 95% prediction interval.