Reevaluation by high-performance liquid chromatography: Clinical significance of microalbuminuria in individuals at high risk of cardiovascular disease in the Heart Outcomes Prevention Evaluation (HOPE) study

Background In the Heart Outcomes Prevention Evaluation Study, microalbumin and albumin levels measured by using radioimmunoassay (RIA) and less than the microalbuminuria threshold in baseline urine samples were associated independently with cardiovascular (CV) events. Conventional immunoassays may underestimate albuminuria by not detecting a mildly denatured unfragmented form of albumin, immunounreactive to conventional antibodies. Methods: Microalbuminuria was reanalyzed in baseline samples stored at -70 degrees C for 5,358 North American participants, 1,992 with diabetes mellitus, by using a high-performance liquid chromatography (HPLC) system that also detects immunochemically nonreactive urinary albumin. Results:The HPLC compared with RIA method identified microalbuminuria in 1,585 versus 719 participants, 809 versus 423 patients with diabetes, by using a conventionally accepted albumin-creatinine ratio (ACR) of 29 mg/g or greater (>= 3 mg/mmol) as a cutoff value. HPLC-detected microalbuminuria increased risk for the primary outcome (a composite of myocardial infarction, stroke, and CV death); unadjusted hazard ratio, 1.85 (95% confidence interval, 1.57 to 2.19). Receiver operating characteristic analysis did not differentiate between HPLC- and RIA-detected microalbuminuria as predictors of CV outcomes. Conclusion: The prevalence of microalbuminuria is 2 to 3 times greater with HPLC than RIA using an ACR of 29 mg/g or greater ( :3 mg/mmol). The optimal cutoff value for detecting CV risk in the entire study population by means of RIA was 9 mg/g (0.9 mg/mmol), and with HPLC, 32 mg/g (3.4 mg/mmol). Results from this study also show different ACR cutoff values for individuals with diabetes: RIA, 13 mg/g or greater (>= 1.4 mg/mmol); HPLC, 44 mg/g or greater (>= 5.2 mg/mmol) and without diabetes: RIA, 7 mg/g or greater (>= 0.7 mg/mmol); HPLC, 29 mg/g or greater (>= 3/1 mg/mmol). Results highlight the importance of method-dependent cutoff values in the prediction of CV events.