2) (BC), apoptosis;
CD95-FITC (clone DX2) (BDB), regulatory T lymphocytes; CD25-ECD (clone B1.49.9) (BC), CD25-FITC (clone B1.49.9) (Immunotech-BC), CD127-FITC (clone eBioRDR5) (eBioscience, San Diego, CA, USA) and DC; HLA-DR- Peridinin-chlorophyll-protein complex (PerCP)-clone L243 (G46-6), Lineage 1 (CD3, CD14, CD16, CD19, CD20 and CD56)-FITC, CD11c-PE (clone S-HCL-3), CD123-PE (clone 9F5) (BDB). Anti-human foxp3-PE (clone PCH101) staining set (eBioscience) was used for intracellular staining of foxp3. The cells were analysed on a Beckman Coulter Cytomics FC 500 MPL flow cytometry equipped with argon and diode laser for five-colour detection. Analyses were performed using mxp version 2.0 (Beckman Selleck NVP-BKM120 Coulter, Sotrastaurin research buy Inc., Brea, CA, USA) flow cytometry software. A gate was set on the lymphocytes according to forward and side scatter properties. Statistical regions were set according to
isotype controls. For foxp3, the statistical marker was set at the upper cut-off for the CD4-negative population following the manufacturer’s instruction. Treg subsets were defined as CD25+/foxp3+ or CD25+/CD127− CD4+ T cells (Fig. 1A–C). DC was analysed for the expression of CD11c and CD123 by gating from HLA-DR+ Lineage (CD3, CD14, CD16, CD19, CD20 and CD56)-negative cells (Fig. 1D–F). Statistical analyses. In a preliminary step, we investigated the data by using histograms and QQ plots for all cell subsets, and computing the Spearman correlations
between all not pairs of cell subsets. This was carried out for the entire data set and for each patient group. Spearman correlations were chosen because of their wider range of detectable relations. Investigating these 12 cell subsets leads to 66 tests, i.e. we have to take into account multiple effects. Because these tests are not independent, the Bonferroni level is too conservative. Thus, we used a significance level of 0.01. The research question contains two different types of comparisons. Comparing the different groups (controls, LTBI and active TB), we used a two-step test procedure. First, we used a Kruskal–Wallis test to detect differences in cell subsets fractions between the groups. In the second step, we selected the cell subsets where the Kruskal–Wallis test detected a significant difference and tested the groups pairwise using a Wilcoxon test to decide where the differences detected by the Kruskal–Wallis test were located. In both cases, we used the Bonferroni significance level, i.e. 0.0042 for Kruskal–Wallis test (12 tests) and 0.0167 for the Wilcoxon test (three tests for each cell subset). Comparing the pre/post-therapy measurements for the QFT+ patients, we used a signed rank test, again with a Bonferroni level of 0.0042. In all investigated cases, we used non-parametric tests because the preliminary analysis indicated a non-Gaussian distribution at least for some of the variables.