Total Testosterone Confounds the Association Between Total Bilirubin and Dyslipidemia

Background: The mechanism for the association between total bilirubin (TBili) and dyslipidemia remains unclear. Total testosterone (TT) has been implicated in reducing bilirubin conjugation and decreasing atherogenic lipids. We hypothesized that 1) TBili was inversely associated with dyslipidemia, and 2) TT confounded this association. Methods: Our study population consisted of 5,878 (2,730 male and 3,148 non-pregnant female) adults aged ≥20 years from the 2011–2016 National Health and Nutrition Examination Survey (NHANES). We excluded those taking self-reported cholesterol medications. Participants with transaminitis (AST or ALT >45 IU/L; AST/ALT >5), excessive alcohol consumption (>20 drinks/week for males; >10 for females), iron overload (transferrin >50%), or positive hepatitis B/C serology were also excluded. We categorized TBili into sex-specific quartiles (Male: <0.5, 0.5–0.6, 0.6–0.8, ≥0.8 mg/dl; Female: <0.4, 0.4–0.5, 0.5–0.6, ≥0.6). Dyslipidemia was defined as elevated TG (≥150 mg/dl) or low HDL (<40 mg/dl for male; <50 for female). We used survey design-adapted multivariable logistic regression, adjusting for TT, demographics, cardiometabolic factors, and liver function. We also stratified by sex-specific median TT levels (386 ng/dl in males; 18.5 ng/dl in females) to determine effect modification. Further, we determined whether the association between TBili and dyslipidemia persisted in males with TT deficiency (<280 ng/dl). Results: Among the 5,878 adults, 1,013 (38%) males & 958 (30%) females had elevated TG, and 803 (29%) males & 1,146 (33%) females had low HDL. Males in the highest quartile (Q4) of TBili had age-adjusted, mean (SD) 50.1 (3.5) mg/dl lower TG and 4.0 (0.9) mg/dl higher HDL than males in the lowest quartile (Q1; p<0.0001). Females in Q4 had 36.4 (4.9) mg/dl lower TG and 5.1 (1.4) mg/dl higher HDL than Q1 (p<0.0001). Males and females in Q4 had 60% and 59% lower odds, respectively, of elevated TG compared to Q1 (adjusted OR [95% CI]; Male: 0.40 [0.28, 0.57], Female: 0.41 [0.32, 0.52]). Males and females in Q4 had 44% and 39% lower odds, respectively, of low HDL compared to Q1 (Male: 0.56 [0.38, 0.81], Female: 0.61 [0.42, 0.90]). Adjusting for TT increased the parameter estimate for Q4, relative to the univariate estimate, by 21% in both sexes. There was no significant difference in TT-stratified odds of elevated TG or low HDL. Among the 544 (19%) males with TT deficiency, Q4 had 56% lower odds of elevated TG and 46% lower, but insignificant, odds of low HDL (aOR [95% CI]; TG: 0.44 [0.21, 0.89], HDL: 0.54 [0.26, 1.12]). Conclusion: TBili was inversely associated with elevated TG and low HDL. TT confounded, but did not modify, this association. Future studies examining TBili’s antiatherogenic role should adjust for TT.