Metabonomics uncovers a reversible proatherogenic lipid profile during infliximab therapy of inflammatory bowel disease

Table 3 Validation of PLS-DA and O-PLS-DA models

Model	PLS-DA permutation test n = 200	O-PLS-DA CV-ANOVA	Area under the ROC curve
CD(0) vs. UC(0)	Q² = 0.095	Q² = –0.182	0.44
CD(0) vs. UC(0)	×	× P = 1	0.44
CD(0) vs. Control	Q² = 0.637, r = 0.32	Q² = 0.7	0.96
CD(0) vs. Control	✓	✓P < 0.001	0.96
UC(0) vs. Control	Q² = 0.583, r = 0.32	Q² = 0.383	0.94
UC(0) vs. Control	✓	✓P < 0.001	0.94
CD Rem(0) vs. Control	Q² = 0.63, r = 0.37	Q² = 0.66	0.96
CD Rem(0) vs. Control	✓	✓P < 0.001	0.96
CD Rem(2) vs. Control	Q² = 0.69, r = 0.37	Q² = 0.66	0.95
CD Rem(2) vs. Control	✓	✓P < 0.001	0.95
CD Rem(6) vs. Control	Q² = 0.52, r = 0.37	Q² = 0.63	0.91
CD Rem(6) vs. Control	✓	✓P < 0.001	0.91
CD Rem(14) vs. Control	Q² = 0.48, r = 0.40	Q² = 0.62	0.90
CD Rem(14) vs. Control	✓	✓P < 0.001	0.90
UC Rem(0) vs. Control	Q² = 0.56, r = 0.47	Q² = 0.60	0.91
UC Rem(0) vs. Control	✓	✓P < 0.001	0.91
UC Rem(2) vs. Control	Q² = 0.60, r = 0.48	Q² = 0.60	0.94
UC Rem(2) vs. Control	✓	✓P < 0.001	0.94
UC Rem(6) vs. Control	Q² = 0.25	Q² = 0.52	0.71
UC Rem(6) vs. Control	×	✓P < 0.001	0.71
UC Rem(14) vs. Control	Q² = 0.17	Q² = 0.37	0.64
UC Rem(14) vs. Control	×	✓P < 0.001	0.64

The models were only considered valid if the permutation test and the CV-ANOVA test (p < 0.05) were satisfied at the same time
(0), before 1st infusion of infliximab; (2), before 2nd infusion; (6), before 3rd infusion; (14), before 4th infusion
Q2, predictability of the model; r correlation coefficient
✓, valid model
X invalid model
CD Crohn’s disease, CV-ANOVA analysis of variance of the cross-validated residuals, O-PLS-DA orthogonal-projection to latent structure discriminant analysis, PLS-DA projection to latent structure-discriminant analysis, Rem remission, ROC receiver operating characteristics, UC ulcerative colitis

ISSN: 1741-7015