Epidemiology call be a useful guide to risk prediction. If, for example, the value of serum
cholesterol,
high density lipoprotein (
HDL) cholesterol and
triglycerides as predictors of future CHD events is considered, then generally the
triglyceride level measured on a single occasion will add little to the prognostic information contained in
cholesterol and
HDL cholesterol. The serum
triglyceride concentration is, however, often more strongly correlated with future CHD incidence in univariate analysis than is serum
cholesterol. However, in multiple logistic regression analysis, particularly when
HDL cholesterol is included, the strength of the apparent independent relationship between
triglycerides and CHD incidence is weakened often to the point of insignificance in individual trials, although it is still evident on meta-analysis of all the epidemiological trials in which both
HDL cholesterol and
triglyceride levels were measured. The erosion of the relationship between
triglycerides and CHD incidence when HDL is included in multiple logistic regression analysis is to some extent is an artefact of the greater
biological variation of
triglyceride concentrations compared with
HDL cholesterol. When allowance is made for this
triglycerides can have more predictive power than HDL. Important clinical decisions are generally not based on single measurements, but on a series which reduces the effect of
biological variation. However, even more importantly epidemiology cannot tell us that lowering
cholesterol or raising
HDL cholesterol levels will have more therapeutic benefit than decreasing
triglyceride levels. That can only be established in clinical trials. An overview of trials involving drugs, which have as their principal action
triglyceride-lowering, revealed them to decrease CHD incidence as much as
statins. In the trials the drugs, which principally lowered
triglycerides, also produced small decreases in serum
cholesterol. The decrease in CHD incidence was, however, more than would be predicted from a similar reduction in
cholesterol achieved with
statins. Epidemiology can thus be a poor guide to clinical decisions. Furthermore the epidemiological relationship between
cholesterol,
triglycerides, HDL and CHD gives us only limited insight into the mechanisms by which these
lipids and
lipoproteins are involved in
atherogenesis and their relative importance in this process. Thus evidence from clinical studies linking hypertriglyceridaemia with potentially important atherogenic factors such as
intermediate density lipoproteins, small dense
LDL and increased
cholesteryl ester exchange may provide a greater understanding of
atherogenesis and potential sites of therapeutic intervention than epidemiology. There is thus evidence for the therapeutic value of lowering
triglycerides and an emerging view that
triglyceride-rich
lipoproteins are frequently crucial in
atherogenesis.