Ira J. Goldberg, MD, is professor of medicine and chief, division of preventive medicine and nutrition, Columbia University College of Physicians and Surgeons, New York, New York.

Conventional measurements of cardiac lipoprotein risk include cholesterol, triglyceride, and high-density lipoprotein (HDL) cholesterol levels. Low-density lipoprotein (LDL) cholesterol is estimated from these measurements. Such measurements, although perhaps not ideal, possess the major advantage of having decades of use leading to large databases from population studies and pharmacologic interventions. A variant on this conventional lipoprotein approach is the use of the total cholesterol/HDL ratio. This is the method used in the Framingham risk calculator that is becoming a standard means for deriving risk-based therapies. New measurements of cardiovascular risk, if widely used, are major commercial ventures. Sometimes, this is worth it. Sometimes, it is not.

The Veterans Affairs High-Density Lipoprotein Intervention Trial (VA-HIT) included a selected population of subjects with a low HDL cholesterol level and limited hypertriglyceridemia. This was a population in which the beneficial effects of gemfibrozil (Lopid) therapy to lower triglycerides without affecting total LDL cholesterol were shown. Why this occurred is still a topic of debate and is discussed by Dr Robins. As a research objective, it is useful to identify, in detail, changes in fasting and postprandial lipoproteins that might relate to the beneficial effects of gemfibrozil in this study. To do this, the investigators collaborated with a commercial company that has developed a novel and widely accepted method to assay lipoprotein size and numbers.

What was learned? Surprisingly to some, measurements of LDL particle size were not predictive of events. LDL particle size measurements by several techniques have become almost routine in some cardiology practices; particle size increases with reduced triglyceride levels, but measuring this might not be useful. The number of LDL particles was predictive of events in this study. Numbers of particles can be determined by an apolipoprotein (apo) B assay; there is only 1 apoB in each LDL and very-low-density lipoprotein (VLDL) particle. One would expect that apoB within LDL, that is, total apoB minus the apoB estimated to be in VLDL, could be obtained in a similar manner to the LDL estimate currently in use. This would also provide LDL particle numbers, but this was not done.

The second predictive index was the number of HDL particles. Because the amount of apoA1 varies on different HDL particles, there is no obvious way to make this estimate by other methods. Not surprisingly, the number of HDL particles was associated with decreased risk. For more than 2 decades, there has been controversy about the “protective” efficacy of different-sized HDL particles. The clinical introduction of cholesterol ester transfer protein inhibitors leading to larger HDL particles and the discovery that 2 different receptors allow reverse cholesterol transport from large and small HDL particles—ABCG1 and ABCA1, respectively—has further complicated this subject. In this study, as in others, more HDL particles are better.

So what is the take-home message for clinicians? Gemfibrozil therapy for patients with high triglyceride and low HDL cholesterol levels is helpful. Although some novel lipid measurements before and during therapy might predict who will benefit from treatment, because there is no dose titration with this drug and no clinical data showing methods to improve benefit from additive therapy, you are at a loss. Perhaps future studies using this type of analysis will help predict those patients with high triglycerides, low HDL cholesterol level, and higher LDL particle numbers. These are the patients who will benefit most by the addition of statins to gemfibrozil therapy, which will reduce LDL particle numbers.