05-06-2017 | Image
(a) Control gut MLGs (≥100 genes; Supplementary Fig. 3) affected by DMARD treatment (P < 0.05, paired Wilcoxon rank-sum test, N = 32). (b) Control dental MLGs (≥100 genes; Fig. 2a) affected by DMARD treatment (P < 0.05, paired Wilcoxon rank-sum test, N = 15). (c) Control dental MLGs differentially enriched in patients with good, moderate or no improvement (N = 9, 21 or 10, respectively; P < 0.05 between good and moderate, good and unimproved, and moderate and unimproved, Wilcoxon rank-sum test). (d) Receiver operating characteristic curve for prediction of improvement after DMARD treatment from before-treatment dental samples. Tenfold cross-validation with a random forest classifier was done five times, and 17 MLGs were selected (Supplementary Table 10). N = 24 for good or moderate improvement, N = 7 for no improvement (Supplementary Table 22). AUC = 0.881; 95% CI shown as a shaded area. The diagonal line marks an AUC of 0.5 (i.e., random classification). (e) Control salivary MLGs (≥100 genes; Fig. 2b) affected by DMARD treatment (P < 0.05, paired Wilcoxon rank-sum test, n = 10). MLGs enriched in control samples are shown in blue text, and MLGs enriched in RA samples are in red text (a–c,e,f). (f) Control dental MLGs differentially enriched in patients after treatment with MTX, MTX + T2 or T2 (N = 14, 10 or 12, respectively; P < 0.05 between MTX and MTX + T2, between MTX and T2, and between MTX + T2 and T2, Wilcoxon rank-sum test). In all box plots, interquartile ranges (IQRs; boxes), medians (dark vertical lines in boxes), the lowest and highest values within a range 1.5 times the IQR from the first and third quartiles (whiskers) and outliers beyond the whiskers (circles) are shown.