Table 1.
Clinical and demographic characteristics of the discovery cohort and the replication cohort at inclusion.
Table 2.
Differentially regulated proteins in Crohn’ disease patients, ulcerative colitis patients, and healthy controls (from univariate analysis).
Fig 1.
Venn diagram of significantly altered proteins (q < 0.05, fold change > 20%) from comparison of UC patients vs. healthy controls (HCs) and CD patients vs. HCs in the discovery cohort.
Seven markers were common between the analyses.
Fig 2.
sPLS plot showing the distribution of all samples from CD patients, UC patients, and healthy controls (HCs) in the discovery cohort, based on the 20 most important proteins from the multivariate analysis.
Fig 3.
sPLS plot showing the distribution of patients with ileal CD, colonic CD, and UC in the discovery cohort, based on the 20 most important proteins from the multivariate analysis.
Table 3.
Cross-validation error rates for discrimination of subgroups of individuals and corresponding resampling p-values.
Fig 4.
Significant biomarkers (q < 0.05, fold change > 20%) from the discovery cohort that passed validation in the replication cohort shown with fold changes and 95% confidence intervals. Actual fold changes from the replication cohort are presented with a red dot in the graph. A: Proteins validated in CD versus HC samples. B: Validated proteins in UC versus HC samples.
Fig 5.
Venn diagram of validated proteins from comparison of UC patients vs. healthy controls (HCs) and CD patients vs. HCs in the replication cohort.
Ten proteins were unique to CD and two were unique to UC.
Fig 6.
Receiver operator characteristics curves (ROC) of the sparse partial least squares models applied on the replication cohort, with area under the curves (AUCs).
L2, colonic Crohn’s disease; L1/L3, ileal/ileocolonic Crohn’s disease; CD in remission, Crohn’s disease patients in clinical remission; UC in remission; ulcerative colitis patients in clinical remission.
Table 4.
Error rates of the sparse partial least-squares model for discrimination of subgroups generated from the discovery cohort, when applied to the replication cohort.