Table 3. Linear regression of anthropometric, biochemical and ceramide data against beta-cell function measures.

Predictor	Beta-cell function/ HOMA-IR (pmol mmol-1)		Disposition index (pmol mmol-1)		Beta-cell function* Matsuda index
	Beta coefficient	P	Beta coefficient	P	Beta coefficient	P
WHR	-0.33	0.001	-0.30	0.002	-0.26	0.016
RA index	-0.24	0.038	-0.25	0.028	-0.25	0.021
Cer 12:1(2H)	-0.24	0.015	-0.24	0.021	-0.23	0.010

Summary of strongest predictors of beta-cell function using linear regression analysis. WHR, waist-to-hip ratio; HDL, high density lipoprotein cholesterol; RA index, resistin-to-adiponectin ratio; cer, ceramide. Data are presented as beta coefficient and P-value according to beta-cell function/HOMA-IR; Homeostatic Model Assessment of Insulin Resistance and DI; Disposition index; beta-cell function (glucose in mg dl^-1, insulin in μIU ml^-1) adjusted for the Matsuda index; P-value determined using backward linear regression analysis. Significance level = P < 0.05. Demographic and Anthropometric variables included were: age, sex, BMI, WHR, BP SYS, BP DIA. Biochemical variables included were: HDL cholesterol, adiponectin, resistin, RA index, triacylglycerides, Apo E, TNFα, IFNγ, IL2, IL4, IL6, IL8, IL10. Ceramide data from lipidomic analysis was examined.

* indicates multiplication.