Whether the protective effect of high density lipoprotein (HDL) on incident coronary
heart disease (CHD) can be attributed to one or both HDL subfractions remains controversial.
The associations of HDL(2) and HDL(3) cholesterol with the incidence of CHD in the
9-year follow-up of the Caerphilly study are described. A total of 2398 middle-aged
British men were recruited from the general population between 1984 and 1988 and were
followed, on average, for 9 years. Total and HDL(3) cholesterol were measured by a
two-step precipitation technique on fresh, fasting samples from 2225 men. HDL(2) cholesterol
was calculated by subtracting HDL(3) from total HDL cholesterol. Relative odds and
95% confidence intervals (CI) for incident CHD were obtained by use of a logistic
regression model. During follow-up, 282 (12%) men developed a major new CHD event.
Total HDL and HDL(3) cholesterol were significantly and inversely associated with
the risk of incident CHD. When divided into fifths of the distributions of total HDL
and HDL(3) cholesterol, multivariate-adjusted relative odds were 1.00, 0.95, 0.72,
0.85, 0.38 and 1.00, 1.05, 0.92, 0.67, 0.39, respectively graded from the least to
the most quintile, with the lowest quintile group as referent. Tests for trend were
significant (P for trend 0.003 and 0.001, respectively). In a multivariate model,
the contribution of HDL(3) was significant (standardized relative odds, 0.76; 95%
CI, 0.64-0.91), whereas HDL(2) was not significant. No linear combination of the two
subfractions was a better predictor of CHD than total HDL cholesterol alone. HDL(3)
cholesterol was an independent predictor of incident CHD and may be more closely related
to the development of CHD than HDL(2) cholesterol. The prediction of the risk of CHD
from total HDL cholesterol alone could not be improved upon by measurement of the
two HDL subfractions. In our view, the only way to improve our understanding of this
situation is to measure both subfractions independently of each other and not to calculate
one by subtraction.