In an interview with Pharmacy Times, Seth Martin, MD, MHS, professor and preventive cardiologist at Johns Hopkins Medicine and director of the Advanced Lipid Disorders Program and Digital Health Lab at the Johns Hopkins Ciccarone Center for the Prevention of Cardiovascular Disease, discussed a new machine learning version of the Martin-Hopkins equation published in JAMA Cardiology and what it means for pharmacists who rely on low-density lipoprotein cholesterol (LDL-C) results to guide treatment decisions.
Key Takeaways for Pharmacists
- The new machine learning version of the Martin-Hopkins equation is a transparent, fixed formula—not a dynamic model—that laboratories can code once and rely on for consistent results.
- Accurate LDL-C directly guides decisions about statins and non-statin therapies, including ezetimibe, bempedoic acid, and PCSK9 inhibitors.
- The 2026 dyslipidemia guideline sets LDL-C goals of less than 100, 70, and 55 mg/dL based on risk status and also includes non-HDL and ApoB goals to confirm therapy optimization.
Martin explained that LDL-C sits at the center of clinical decision-making around therapies that reduce atherosclerotic cardiovascular disease (ASCVD) events and mortality, making accuracy a matter of life or death for patients. The new machine learning version, he emphasized, is not a dynamic model that shifts over time but a transparent, fixed equation that laboratories can code once and rely on year after year. It produces results similar in accuracy to the original Martin-Hopkins equation, published more than a decade ago, while making it easier for laboratories to automate the calculation so that pharmacists and other busy clinicians receive trustworthy results without additional steps.
The core clinical problem, Martin said, is that the Friedewald equation and other commonly used formulas tend to underestimate LDL cholesterol, producing falsely reassuring results. A patient whose LDL appears low may in reality be 10 to 20 mg/dL higher and above the threshold that would qualify them for evidence-based lipid-lowering therapy, including statins and non-statin options such as ezetimibe (Zetia; Organon), bempedoic acid, and proprotein convertase subtilisin/kexin 9 (PCSK9) inhibitors.
Martin described the "stress test" of LDL accuracy as patients with lower LDL levels and higher triglycerides—for example, a patient with very high-risk ASCVD whose LDL returns near 50 mg/dL with triglycerides in the 200 to 300 mg/dL range. In that scenario, the Friedewald or Sampson-NIH equations may show a patient at goal when the Martin-Hopkins equations would place them above it.
He urged pharmacists to confirm which equation their local laboratory uses, to serve as local champions for adopting more accurate calculations, and to keep the 2026 dyslipidemia guideline's LDL goals of less than 100, 70, and 55 mg/dL in view alongside non-high-density lipoprotein cholesterol and apoliprotein B targets.
