Pressure by Oxfordshire GP’s for screening facilities for high density lipoprotein (HDL) is a welcome development. Unfortunately the predictive ability of HDL alone is not however accurate enough in coronary heart disease (CHD) risk estimation. True an inverse relation between HDL and CHD has been confirmed by prospective studies and subsequently shown to be independent of other risk factors.
The usual HDL blood level is between 1.1 and 1.7 mmols with raised levels found in athletes and pre-menopausal women – higher than average levels regarded as protective against CHD. Levels below 1.0 are predictive of increased susceptibility to CHD. HDL levels relate negatively to degree of obesity and show a positive correlation with exercise participation, physical fitness and alcohol consumption. Plasma HDL tends to be low amongst the over-weight, sedentary individuals and smokers. The Kilkenny Health Study (1985) detected amongst males and females the following HDL levels:
Age Male Female
35-44 0.91 1.13
45-54 0.90 1.13
55-64 0.94 1.13
Age-sex related HDL levels.
Kilkenny Heart Study, 1985.
The Heartbeat Wales Survey (1985) showed that HDL levels fall with increasing age, with female levels higher at all ages. An Israeli study (1985) showed that age adjusted mortality increased with decreasing HDL levels. HDL therefore provides an additional tool for assessment of CHD risk with, according to the British Regional Heart Study (Pocock, S. J. 1986), a one standard deviation increase in HDL indicating a 10% decrease in risk, whereas a one standard deviation increase in non-HDL (TC-HDL) indicating a 50% increase in risk. Further estimations (Pocock, S. J. 1989) calculated a 2.5% reduction in IHD with every 10mg/L (0.026 mmol) of HDL elevation.
Most importantly though determination of HDL enables us to calculate the low density lipoprotein (LDL) component – the major atherogenic particle of total cholesterol (TC). Measurements of these cholesterol fractions can be found by knowing two of them from the following formula: LDL = TC – TG/5 – HDL (in mmols). This is useful because studies of CHD risk assessment emphasises the predictive power of the LDL:HDL ratio. LDL can be calculated using the Friedwald Formula (FF) of TC – HDL – (TG x 0.45) in mmols or in another form: LDL = TC – HDL – TG/2.19 (mmols).
Does estimation of HDL cholesterol subfractions imply the impending redundancy of serum TC or other lipoprotein fractions? The answer must be no. Even though serum TC is only a first approximation requiring serial validation a generalised TC screening procedure is still required. This is because any population strategy for CHD risk estimation must encompass the determination of as complete a lipid profile as possible.
Serum triglycerides (TG) have been shown to predict IHD risk in men with low serum TC (= < 220 mg/100 mls) and the Framingham Study viewed TG as a risk factor if the TC to HDL ratio exceeded 3.5. In this respect calculation of LDL is more precise for plasma with TG concentrations within normal range using De Long’s modification of the original FF thus: LDL = TC – (HDL + 0.37 x TG) for mmols (0.16 x TG for mg/100 mls). In order to calculate LDL using an FF that varies with TG it is better to use Rao’s modification thus: LDL = TC – HDL -TG (0,47 – (0.022 x TG). LDL estimates of risk based on recent transatlantic concensus suggest that with hypercholesterolaemia as the only risk factor it is advisable to reduce LDL to below 4.9 mmol. With hypercholesterolaemia plus two other risk facors an LDL level of 4.1 mmol was recommended.
Hyperlipidaemia has been described as serum TC > 7.2 mmol and/or serum TG > 2.0 mmol – though caution has to be exercised here because of population variation. For example the estimated mean TC in eastern Finland (a high CHD country) is 270mg/100 mls (6.98 mmol), whereas Japan (a low CHD country) has an estimated mean TC of only 170mg/mls (4.39 mmol). In addition, even though there is disagreement about TG’s as independent risk factors for CHD, hypertriglyceridaemia in the presence of low HDL concentration is possibly associated with increased vascular disease (hypertriglyceridaemia = > 2.00 mmol, HDL = < 0.9 mmol).
The complexity of inter-population and intra-population variation in lipid profiles is further compounded by intra-individual variation. For example it has been shown that TC shows temporal variations over a six hour period – from a nadir of 189 mg/100 ml (4.26 mmol) between 10 and 11 am to a zenith of 211 mg/100 ml (5.45 mmol) between 3 and 4 pm, with a periodic cycle of 50 minutes. How often are samples taken in surgeries and clinics during morning hours? Such diurnal and circadian variations shown by lipid levels must be noted when determining lipid profiles in both the individual patient and in populations.
Despite the arbitrariness of cut-points guidelines must remain for assessment of serum lipid parameters. The National Institutes of Health (USA) determined the following groups in mmols:
Age Acceptable Moderate Risk High Risk
20-29 > 5.17 5.18 – 5.68 < 5.69
30-39 > 5.69 5.10 – 6.20 < 6.21
40 + > 6.21 6.22 – 6.71 < 6.72
National Institutes of Health, USA (in mmols).
MONICA (monitoring trends in coronary heart disease study) in Scotland settled on the following cut-points by age and sex for median TC levels in 10 year groups, in mmols:
Age Men Women
23-34 5.5 5.2
35-44 6.0 5.5
45-54 6.3 6.4
55-64 6.2 7.2
MONICA, Scotland
10 year cut-points by sex for median TC.
The European Atherosclerosis Society guidelines suggested 5.2 mmol as acceptable. 5.2 to 6.5 mmol required dietary intervention, 6.5 to 7.8 mmol required lipid lowering agents and < 7.8 mmol needed referral to a specialist lipid clinic. The British Four Centres Sudy (Mann, J. I. 1988) used cut-points for TC of 5.5, 6.5, and 8.0 mmol with over 6.5 indicating clinical care – the mean male and female blood TC in Oxford of 5.8 mmol were above the World Health Organisation optimum level of 5.17 mmol.
High risk emerges above the 75th percentile for TC and accounts for 40 to 50% of CHD cases, but 50 to 60% of CHD cases appear in the 75% of the population below the 75th percentile (with familial hypercholesterolaemia at the 99th perecentile) – and obvious case for more accurate lipid analysis that serum TC measurements that ignore HDL and the atherogenic LDL component. For example a case study from the Cleveland Clinic Foundation of a woman athlete with elevated TC was considered as a moderate risk using 75th percentile measurements. Further tests showed an elevated HDL that reduced her LDL to safe limits. Again this emphasises the validity of partitioning TC into atherogenic LDL and anti-atherogenic HDL fractions.
In addition, estimation of CHD susceptibility utilising serum cholesterol presents problems because the relation of TC to CHD incidence becomes attenuated with advancing age. The Framingham Study has shown that beyond the age 55 TC levels can no longer predict CHD. However, with fractionation into lipoprotein components (LDL, HDL, VLDL) the relationship of the lipid profile to CHD re-emerges. This is particularly of value in assessing risk in those individuals susceptible to premature CHD in age groups 45 to 64. Likelihood ratios for men (TC 1.98, HDL -14.0, LDL 4.4, VLDL 1.03) and women (TC 2.26, HDL -21.2, LDL 4.5, VLDL 5.04) represent predictive strengths for the 50 to 80 years group, A number of ratios for likelihood risk for both sexes aged 50 to 80, that include LDL:HDL ratios, have been calculated for expectancy in both normo- and hyerlipidaemic individuals.
A valuable predictive model once HDL and LDL values are known is the estimation of the Cholesterol Retention Fraction (CRF) used in the Bowling Green Study (USA). This formula has not, as far as is known, ever been applied to any British data. The only use of CRF in Britain at present is being applied, by the present author and researcher, to data from the Heartbeat Wales Study to which the author has access. The CRF is derived from (LDL – HDL)/LDL which gives the following predictor values:
Predictor values
> 0.70 ———-> atheroscerosis
< 0.69 ———-> mitigates CHD
< 0.60 ———-> implies immunity
CRF applied to Heartbeat Wales data.
The accuracy of CRF depends upon the proviso that LDL concentration is < 4.4 mmol or LDL is 1.05 mmol and that systolic blood pressure is < 140 mm Hg. Here we have lipid predictors linked to BP – a much more enhanced model for individual and population estimation of CHD risk and susceptibility.
HDL assessment should be part of a wider population prevention strategy, an approach that attempts to determine population attributable risks for a number of risk factors in all age groups – especially the young. This was the main aim of the ill-fated proposed project titled the Oxford City Heart Health Initiative which received no applied for national or local funding. The estimation of HDL does not provide a key to open all locks, but it does give GP’s a more precise tool to estimate lipid profiles and correlate them with other risk factors for CHD susceptibility in their patients.
[An unpublished reply to an article in the Oxford Times. Submitted 30.6.1990, while author was engaged on university research into the anthropology and genetics of CHD risk factors including HDL, LDL and total cholesterol].
Eric Edwards Collected Works 