Table 2.

The results from the OPLS-DA and statistical analyses

OPLS-DA model diagnostics
Predictive component			R2X	R2Y	Q2
			0.0995	0.477	0.28
Orthogonal components			R2X(o)
			0.548
			0.107
cv-ANOVA p value			0.0001
Misclassification table
	Members		Correct
preCHT	1		100%
postCHT	23		100%
List of the important metabolites from the OPLS-DA model
Name		ppm	p(corr)	p value	Median ratio [%]
Metabolites increased after induction chemotherapy
1	Lipids	0.9	0.76	< 0.000	12.73
		1.3	0.57	< 0.000	18.58
		5.3	0.69	< 0.000	17.38
Metabolites decreased after induction chemotherapy
2	Alanine	1.48	0.33	0.0008	14.21
3	NAG	2.07	0.12	0.0002	9.03
4	Glucose	3.24	0.31	0.04	9.4
		3.42	0.43	0.042	8.11
		3.44	0.48	0.038	8.46
		3.51	0.43	0.036	9.54
		3.56	0.46	0.04	9.33
		3.72	0.55	0.039	9.59
		3.76	0.43	0.031	8.3
		3.83	0.39	0.019	8.57
		3.9	0.48	0.028	9.34
		5.2	0.32	0.025	9.57

R2X — an amount of variation in the data that is correlated to class separation; R2Y — a fraction of the class membership (Y) variation modeled using the data matrix (X), this parameter tells how good is the separation between two classes; R2X(o) — an amount of variation in the data that is uncorrelated (orthogonal) to the class separation; Q2 — a predictive ability of the OPLS-DA model; p(corr) — describes reliability of a variable, the closer to one the better; p value – from the WSR test, Median ratio — shows the between class differences in the peak integrals and is calculated as: 100 - (lower median)/(higher median)*100. NAG — N-acetyl-glycoprotein.