We measured the precision of Bio-Plex and MILLIPLEX in quantifying spiked cytokine recovery across repeats of biological replicates within each individual assay, which we report as repeatability. Four identical aliquots of three different patient samples were included at different positions on the same plate. The coefficient of variation (%CV) was calculated for each sample and a mean

%CV derived from the MAPK Inhibitor Library molecular weight pooled %CV values. In this analysis the %CV was lower with the MILLIPLEX kit for IFNγ (15.4% vs 39.3%) and with the Bio-Plex kit for IL-17 (15.6% vs 21.7%). We also measured the intra-assay precision of these two kits in quantifying cytokine concentrations derived from and included in standard curve calculations. The pooled mean %CV across all IL-17 standards was lower with the Bio-Plex kit (11.8% vs 24.2%) and across all IFNγ standards was lower with the MILLIPLEX kit (14.2% vs 25.1%). We have insufficient data to report on inter-assay precision. Complex

biological samples derived from tissues have not been evaluated by Luminex kit manufacturers and the optimal procedure to prepare our human mucosal tissue samples was not known. Determining the impact of different protocols on cytokine measures could improve the utility of Luminex-based methods to achieve our intended purpose — namely the quantification of endogenous cytokines present at low concentrations in small tissue samples. We compared processing methods and extraction buffers for four pairs of biopsies from each of four patients. Within each pair, biopsies were spiked at 100 pg/mL or spiked with buffer alone (“unspiked”), Afatinib cell line processed and then split into aliquots. Dynein Manual sample disruption using a mini pellet pestle with or without homogenisation using a needle and syringe, and automated processing using a TissueLyser LT bead-basher (QIAGEN) were compared, as detailed in Materials and methods. Cytokine spikes were recovered significantly more accurately from

samples processed manually (Fig. 1C). There were no significant differences between processing methods in relation to precision (data not shown) or total protein recovery by BCA assay (mean ± SD for manual 821.8 ± 108.0 μg/mL vs automated 800.3 ± 179.2 μg/mL). We compared manual disruption using pestle alone with additional homogenisation using needle and syringe. Spiked cytokine recovery was usually lower with the latter (Table 2), although this difference was not consistent or statistically significant. We observed that homogenisation with a needle and syringe leads to loss of sample volume, which was retained in equipment dead space. In addition we evaluated if the addition of benzonase to PBS-based extraction buffer improved the performance of manual or automated processing. Benzonase is an endonuclease and digestion of nucleic acids may reduce sample viscosity.