Low-density lipoprotein cholesterol (LDL-C) is used to assess Atherosclerotic Cardiovascular Disease (ASCVD) risk and to manage lipid-lowering therapy. Proprotein convertase subtilisin/kexin (type 9) serine protease (PCSK9) inhibitors are an emerging therapeutic option for decreasing LDL-C levels. Studies have shown that PCSK9 inhibitors can decrease LDL-C by up to 70% when used in conjunction with statins; however, because of its high cost, insurance companies restrict its use to when it is only medically necessary. To be eligible for PCSK9 therapy, a patient has to have pre-existing ASCVD and an LDL-C greater than 70 mg/dL, while on a maximal tolerated dose of statins. Most clinical laboratories calculate LDL-C using the Friedewald Equation (FWLDL-C), which utilizes results from the standard lipid panel (total cholesterol (TC), triglycerides (TG), and high-density lipoprotein cholesterol (HDL-C)). However, this equation often leads to an underestimation of LDL-C, especially when TG levels are elevated. Other more accurate LDL-C equations have recently been developed, such as the Martin (MLDL-C), Sampson-NIH (SLDL-C) and extended Martin (ext-MLDL-C) equations. Using LDL-C determined by the Beta-quantification reference method (BQ), we developed an enhanced version of the Sampson-NIH Equation (eSLDL-C), which also uses apoB, to predict LDL-C.

To test the accuracy of the various equations, we compared BQLDL-C to FLDL-C, MLDL-C, extMLDL-C, SLDL-C and eSLDL-C by regression analysis. We then compared the accuracy of the various equations to the BQ reference method for classifying patients as either being below or above the 70 mg/dL treatment decision threshold for PCSK9 therapy. The eSLDL-C equation performed substantially better than the other equations, with a lower mean absolute difference (MAD) than any of the other equations, indicating that the average difference between the estimated LDL-C and the average BQ LDL-C level was lowest with the eSLDL-C equation. The eSLDL-C equation also had the best overall normalized Matthew’s Correlation Coefficient (nMCC) for identifying patients with a true LDL-C greater than 70 mg/dL as determined by the BQ reference method. The nMCC is a statistical method that only gives a high score if there are good results in all confusion matrix categories (true positives, false negatives, true negatives, false positives). Out of 1125 patients identified to have an LDL-C <70 mg/dL in our dataset by the BQ reference method, approximately 131 of them would falsely still be eligible for PCSK9 therapy because of a positive bias, whereas only 88 patients would have a falsely elevated LDL-C >70 mg/dL by the eSLDL-C equation. Of the 553 patients that are falsely denied by FLDLC, 83% (461 of them) are eligible by eSLDLC, 61% (339) by eMLDLC and 44% (242) by SLDLC.

Here, we show that our enhanced Sampson-NIH equation, which includes apoB for estimating LDL-C levels, more accurately predicts low LDL-C values than other LDL-C equations. The use of the new equation would improve access to those patients who could benefit from the therapy but have falsely low LDL-C by the FWLDL-C equation, which occurs in about half of these patients. Our equation reduces the occurrence of falsely low LDL-C as seen in the FWLDL-C equation. The new equation is also more specific than the MLDL-C and eMLDL-C equations and would reduce the overutilization of PCSK9 therapy for those patients who do not require it, according to current guidelines.