Table 4. Prediction of IGT and T2D in the KORA cohort.

Model	Glycine	LPC (18:2)	C2	Glycine, LPC (18:2), C2
(A) Metabolite as continuous variable (n=589)
Per s.d.	0.75 (0.58–0.95)	0.72 (0.54–0.93)	0.92 (0.73–1.14)	0.36 (0.20–0.67)
P	0.02	0.02	0.50	0.001
(B) Metabolite as categorical variable (n=589)
First quartile	1.0 (reference)	1.0 (reference)	1.0 (reference)	1.0 (reference)
Second quartile	1.0 (0.80–1.46)	0.96 (0.73–1.27)	0.89 (0.66–1.23)	0.54 (0.30–0.97)
Third quartile	1.0 (0.74–1.34)	0.71 (0.51–0.99)	0.93 (0.69–1.26)	0.66 (0.37–1.18)
Fourth quartile	0.78 (0.55–1.06)	0.78 (0.54–1.12)	0.99 (0.73–1.35)	0.36 (0.19–0.69)
P for trend	0.06	0.05	0.79	0.0082
(C) Metabolite as continuous variable (n=876)
Per s.d.	0.73 (0.55–0.97)	0.70 (0.51–0.94)	0.94 (0.74–1.18)	0.39 (0.21–0.71)
P	0.04	0.02	0.59	0.0002
(D) Metabolite as categorical variable (n=876)
1st quartile	1.0 (reference)	1.0 (reference)	1.0 (reference)	1.0 (reference)
2nd quartile	0.87 (0.71–1.07)	0.95 (0.77–1.17)	1.05 (0.85–1.31)	0.50 (0.33–0.76)
3rd quartile	0.82 (0.67–1.01)	0.70 (0.56–0.88)	0.97 (0.78–1.19)	0.57 (0.38–0.88)
4th quartile	0.67 (0.54–0.84)	0.68 (0.54–0.88)	1.21 (0.98–1.50)	0.33 (0.21–0.52)
P for trend	0.00061	0.00021	0.19	1.8E−05
(E) Linear regression (n=843)
β Estimatesa (95% CI)	−2.47 (−4.64, −0.29)	−4.57 (−6.90, −2.24)	1.02 (−1.11, 3.15)	−4.23 (−6.52, −2.31)
P	0.026	0.00013	0.59	8.8E−05

Odds ratios (ORs, 95% confidence intervals) and P-values of multivariate logistic regression results are shown in (A) and (B) for IGT and in (C) and (D) for T2D, respectively, whereas β estimates and P-values from linear regression analysis between metabolite concentration in baseline KORA S4 and 2-h glucose values in follow-up KORA F4 are shown in (E). All models were adjusted for age, sex, BMI, physical activity, alcohol intake, smoking, systolic BP and HDL cholesterol.

^aβ Estimate indicates the future difference in the glucose tolerance corresponding to the one s.d. differences in the normalized baseline metabolite concentration.