The two-sided two-sample t-test was used to compare the mean change in TGF-β between the treatment group and the control group. The significance level was 0·05. For secondary outcomes, the change from baseline (day 0) to values at days 3, 14, 28 and 63 was compared by group. Secondary measurements included expression of CD26 on lymphocyte subsets in PBMCs, percentages of lymphocyte subsets within PBMCs, cytokine and chemokine concentrations in plasma, cytokine

and chemokine Sotrastaurin in vivo concentrations in LPS-stimulated PBMCs, clinical complete blood count (CBC) values, gene expression in whole blood, proliferation and production of cytokines and chemokines (including TGF-β) in supernatants from anti-CD3-stimulated PBMCs. Comparison of the two groups at specific time-points was performed with t-test or Mann–Whitney

U-test for quantitative variables and χ2 or Fisher’s exact test for categorical variables. The primary analysis of these secondary variables included day 28 only, and did not adjust for baseline. Subsequent analyses used generalized linear models to investigate changes over time and a Bonferroni’s correction was applied to account for the multiple time-points. A P-value of 0·0125 was used. These analyses included an adjustment for baseline and log transformation GSK2118436 as needed. The analysis of correlations between the change in activity level of DPP-4 and changes

in immune parameters was performed using Pearson’s correlations using GraphPad Prism. Each individual’s percentage change in DPP-4 activity level Niclosamide was calculated from their individual day 0 value and each subsequent on-drug time-point (days 3, 14, 28); Pearson’s correlations were then calculated between this percentage baseline DPP4 activity and immune parameters (calculated as change from baseline at each time-point, as indicated above). A significant increase in active GLP-1 levels was observed in the sitagliptin group but not the placebo group, indicating that this group was taking active drug (Fig. 2a). As expected, because participants were fasting at the time of the blood draws, GLP-1 levels were low, with an average of 4·9 pg/ml at baseline (day 0). In addition, DPP-4 enzyme activity levels were measured, and a significant drop (P < 0·0001) in the percentage activity compared to day 0 was observed in the sitagliptin group, but not the placebo group (Fig. 2b). On average, while taking sitagliptin, this group showed 50–60% inhibition of activity. The primary outcome for this study was the change in total plasma TGF-β levels from baseline (day 0) to day 28, comparing the group that received sitagliptin and the placebo group.