Performance of Methods to Estimate Low-Density Lipoprotein Cholesterol in Women With and Without HIV Infection

Background Low-density lipoprotein cholesterol (LDL-C) is estimated from total cholesterol, high-density lipoprotein cholesterol and triglycerides using predefined equations which assume fixed or varying relationships between these parameters and may underestimate or overestimate LDL-C. Data on the performance of these equations in persons with HIV are limited. We sought to investigate the utility of the 3 most widely used methods (Friedewald, Hopkins, and the recently proposed NIH equation) to predict LDL-C in persons with HIV. Methods We identified 7397 direct LDL-C (5219 HIV, 2127 uninfected controls, 51 seroconvertors) measurements in the Women's Interagency HIV Study, and used the 3 equations (Friedewald, Hopkins, and NIH) to calculate LDL-C. We compared the performance of the 3 equations using root mean square error and coefficient of determination (R2). Results Overall, the Friedewald equation had the best performance characteristics, outperforming Hopkins and NIH methods with lower root mean square error and higher R2 at lower triglyceride levels. However, this association did not hold true at higher triglyceride levels (quartiles 3 and 4), whereas the Hopkins equation had better performance characteristics in quartile 3, none of the 3 equations were optimal in quartile 4. After adjusting for fasting status and triglycerides levels, HIV+ had larger mean difference compared with directly measured LDL using all 3 methods. Conclusions All 3 methods have lower accuracy in HIV+ vs HIV- women, even after adjusting for triglyceride levels and fasting status. Further research should focus on identifying methods to estimate LDL-C in HIV.