Predictive value of random sample urine bile acids corrected by creatinine in liver disease

Bile acids, in a random sample of urine, discriminated normal controls from liver disease, with a probability similar to fasting plasma bile acids (p less than 0.01 and p less than 0.001, depending on the analytical technique). A high degree of correlation between urinary and plasma bile acids (up to r = 0.93) was achieved only when the urine flow was corrected by using a urinary bile acids/creatinine ratio but not with urinary bile acids as simple volume concentration. These findings originated from 10 patients with severe liver disease and 10 with mild liver disease, all biopsy-confirmed, compared with controls. In 24 normal controls and 24 histologically confirmed compensated liver disease, the predictive value of urinary bile acids/creatinine equaled or exceeded fasting plasma bile acids and routine liver tests. In a patient recovering from subacute hepatic necrosis, the urinary bile acids/creatinine closely resembled changes in plasma bile acids and in the routine liver tests. When sulfated urinary bile acids were included, the discrimination between liver disease and controls did not improve. Gallbladder contraction induced by parenteral analogs of cholecystokinin did not change urinary bile acids/creatinine, despite a significant increase in the plasma bile acids. Collection of fasting urine is thus not necessary. Urinary bile acids/creatinine in 12 subjects with renal insufficiency and moderate impairment of creatinine clearance was not different from controls. The weight/height index did not affect this urinary test: there was no significant correlation between the two. Available radioimmunoassays for plasma bile acids can be easily adapted for urine.(ABSTRACT TRUNCATED AT 250 WORDS)