The detection limit of the granzyme B assay was determined as the lowest amount of granzyme B which could still be detected in the lysate [33]. Per laboratory, an average limit of detection was determined learn more from 12 different assays. The limit of detection was assigned with minor changes from the ICH guideline (33) as 3.3 standard deviations above the lowest amount of granzyme B detectable in the assay.

Precision (consisting of repeatability, intermediate precision, and/or reproducibility) of the granzyme B assay and multiplex assay was determined by replicate analysis of the bulk lysate or supernatant, respectively. Robustness was determined by replicate stimulations of PBMC aliquots from two representative donors with high and low cellular responses to influenza, respectively. The two donors were selected in pilot experiments using the granzyme B

and cytokine assay for determination influenza-specific cellular responses. All essential materials, including frozen PBMC from the selected donors, the bulk lysates and supernatants together with reagents required for the stimulation experiments (mock, H3N2, Con A, human serum), were shipped on solid CO2 to the participating laboratories by express mail. The participants were requested to test these according to the protocols as described. Laboratory personnel who were not experienced with the assays were first trained in a three-day course before starting with the validation program. Statistical analysis was performed using Excell and GraphPad Prism software version 4.03. For verification of Vandetanib chemical structure normal distribution of data Q–Q plots and Kolmogorov–Smirnoff tests were performed applying the SPSS 12.0.1 statistical program. Coefficient of variation (CV), in percentages, was calculated by standard deviation/mean × 100%. Polynomial regression of the standard line showed a correlation coefficient >0.99 in the range of 0–20 granzyme B units (Fig. 2a). Granzyme B

levels ranged between 0.6 and 1.3 units after mock stimulation of PBMC, between 1.3 and 7.5 units after H3N2 stimulation and between 7.5 and 20 units after Con A stimulation, respectively. For each laboratory, granzyme B amounts above the respective detection limit were included in the results. The average detection limit isothipendyl of all laboratories was 0.076 with a CV of 25% (data not shown). To determine whether the granzyme B assay could specifically and accurately measure granzyme B content, lysate derived from PBMC stimulated with Con A was diluted and spiked with 10 units of recombinant granzyme B (Table 1). Samples above the quantitation limit showed a recovery ranging from 94% to 108% which is within the acceptable range for a specific and accurate assay [34] and [35]. Precision of the granzyme B assay was determined by four laboratories from different countries using lysates derived from one batch of PBMC stimulated with mock, H3N2, or Con A (Fig. 2b).