Skip to main content

05-06-2017 | Image

Figure 5: Gut and oral MLGs can be used to distinguish RA patients from healthy controls.

(a,d,f) Receiver operating characteristic curves for fecal (a), dental (d) and salivary (f) training sets comprising samples from treatment-naive RA subjects and unrelated controls (N = 157, 100 and 94 for fecal, dental and salivary samples, respectively). AUC = 0.9396 for fecal, 0.8702 for dental and 0.8135 for salivary samples. The 95% confidence intervals (CIs) are shown as shaded areas. (b) Classification of fecal samples from 17 controls and 17 RA subjects, either consanguineous or nonconsanguineous relatives. Open circles, controls; filled circles, RA subjects. (c,e,g) Classification of fecal (c), dental (e) and salivary (g) samples from DMARD-treated RA patients (N = 40, 37 and 24 for fecal, dental and salivary samples, respectively), shaded on a scale relative to DAS28. NA (no shading), DAS28 not available. The classification results for all samples are listed in Supplementary Table 1. Diagonal lines in graphs mark an AUC of 0.5 (i.e., random classification). Horizontal lines mark the probability cutoff (0.5).