Marja-Riitta Taskinen1; Christian Ehnholm2; David Sullivan3; Kristy Mann4; James D Best5; R J Simes6; Philip Barter7; Anthony C Keech8
　　1 Helsinki Univ Central Hosp and Biomedicum， Helsinki， Finland
　　2 Biomedicum Helsinki， Helsinki， Finland
　　3 Royal Prince Alfred Hosp， Sydney， Australia
　　4 NHMRC Clinical Trials Cntr， Univ of Sydney， Sydney， Australia
　　5 Univ of Melbourne， Sydney， Australia
　　6 NHMRC Clinical Trials Cntr， Univ of Sydney， Sydney， Australia
　　7 Univ of Sydney， Sydney， Australia
　　8 NHMRC Clinical Trials Cntr， Univ of Sydney， Sydney， Australia; on behalf of the FIELD study investigators， c/o NHMRC Clinical Trials Cntr， Univ of Sydney
　　Background. The apo B/apo A-I ratio may be a better indicator of cardiovascular risk in people with type 2 diabetes than is captured by the traditional lipid risk markers (LDL-C， HDL-C and triglyceride). However， using the apo B/apo A-I ratio as a target for lipid-lowering therapy is still debated， with limited data from people with type 2 diabetes.
　　Objective. We compared the prognostic value of apolipoprotein B， apo A-I， apo A-II and the apo B/apo A-I ratio with that of traditional lipid parameters in predicting CVD risk.
　　Methods. The FIELD study randomized 9795 patients with type 2 diabetes to fenofibrate (200 mg daily) or to placebo and followed them up for a median of 5 years. We estimated the hazard ratio (HR) of the effect of 1 SD difference in baseline concentrations of lipids， apolipoproteins， and ratios on the risk of CVD events.
　　Results. There were 683 CVD events in the placebo group and 612 in the fenofibrate group， a significant reduction of 11% by fenofibrate (HR 0.89， 95% Cl 0.80 – 0.99， P = 0.035). The 4 variables best predicting CVD events were， in order， non-HDL-C/HDL-C， total cholesterol/HDL-C， apo B/apo A-I， and LDL-C/HDL-C ratios in the placebo group; in the fenofibrate group apoB/apo A-1 was fourth (table). The HRs were 0.83 for HDL-C， 0.84 for apo A-I， 1.16 for LDL-C (all P < 0.001)， 1.14 for non-HDL-C and apo B (both P < 0.001) in the placebo group， and 0.84， 0.87 (both P < 0.005)， 1.07 (P = 0.09)， 1.22 and 1.22 (both P < 0.001)， respectively， in the fenofibrate group. Apo A-II was nonsignificant (HR 0.94， 95% CI 0.86 –1.02) in the placebo group and borderline in the fenofibrate group (HR 0.92， 95% CI 0.84 –1.00， P = 0.045).
　　Conclusion. As a determinant of CVD risk， the apo B/apo A-I ratio did not perform better than non-HDL-C/HDL-C and total cholesterol/HDL-C ratios. HDL-C and apo A-I in isolation were similarly predictive. Replacement of traditional lipid values with apo B， apo A-I， and apo A-II added little to CVD risk assessment in these patients with type 2 diabetes.
　　Adjusted risk* of CVD events， for the 4 best predictors in the model， by rank in the placebo group