Drugs 40 (Suppl. I): 7-12, 1990 0012-6667/90/0 I 00-0007/$3.00/0 © Adis International Limited All rights reserved. DRSUP17538
HDL-Cholesterol as a Risk Factor in Coronary Heart Disease An Update of the Helsinki Heart Study
M.H. Frick, V. Manninen, 1.K. Huttunen, OP. Heinonen,
L. Tenkanen and M. Miinttiiri
First Department of Medicine, University Central Hospital and National Public Health Institute, Helsinki, Finland
Summary
The aim of the Helsinki Heart Study, a 5-year primary prevention placebo-controlled study involving 4081 dyslipidaemic men (aged 40 to 55 years), was to investigate if increasing high density lipoprotein (HDL)-cholesterol plasma levels and decreasing low density lipoprotein (LDL)-cholesterollevels would reduce the incidence of coronary heart disease. Gemfibrozil 600mg twice daily was administered to induce these changes in lipoprotein levels. Baseline HDL-cholesterollevels in the study group were similar to those in the general population. Data from patients treated with placebo were analysed to investigate the influence of HDL-cholesterol levels on the incidence of coronary heart disease. Using the number of cardiac end-points per 1000 person-years to indicate the risk of coronary heart disease, it was clear that elevated HDL-cholesterol levels reduced the risk of coronary heart disease while the incidence increased at low HDL-cholesterol levels. This relationship was not altered when the effect of HDL-cholesterollevels was analysed jointly with other coronary risk factors (age, smoking or blood pressure). A weaker association was seen between LDL-cholesterol and risk of coronary heart disease, and triglycerides appeared to have no significant effect on the incidence of the disease. The data clearly suggest that HDL-cholesterol is a strong predictor of the incidence of coronary heart disease in the placebo group of the Helsinki Heart Study.
Evidence from large scale epidemiological surveys and controlled clinical trials positively links low density lipoprotein (LDL)-cholesterol to the atherogenic process in different vascular regions, notably the coronary arteries (Martin et al. 1986; Ross 1986). The primary evidence for the protective effect of high density lipoprotein (HDL)-cholesterol was found some 35 years ago (Barr et al. 1951; Nikkilii 1953) and has been strengthened by
several epidemiological studies (Abbot et al. 1988; Miller & Miller 1975). Evidence from studies using coronary angiography is also in line with this reasoning (Berg et al. 1975; Pearson et al. 1979). This paper analyses the impact of HDL-cholesterol on the incidence of coronary heart disease in a 5-year follow-up study of dyslipidaemic, middle-aged males in the Helsinki Heart Study (Frick et al. 1987; Manninen et al. 1988; Miinttiiri et al. .1987).
8
Drugs 40 (Suppi. 1) 1990
1. Materials and Methods
40
The Helsinki Heart Study was conducted as a double-blind, placebo-controlled primary prevention study to assess the effects of gemfibrozil 600mg administered twice daily in middle-aged (40 to 55 years) dyslipidaemic men. The design and primary results have been published previously (Frick et al. 1987; Manninen et al. 1988; Manttari et al. 1987). Briefly, the basic aim of the study was to test the hypothesis that the incidence of coronary heart disease can be reduced by increasing the plasma level of HDL-cholesterol, while simultaneously decreasing LDL-cholesterol levels. Gemfibrozil was chosen for this long term (5-year) study since it is well tolerated after continued use (Olsson et al. 1976) and is an effective lipid modifier. The drug increases HDL-cholesterol and decreases total and LDL-cholesterol levels in healthy volunteers and in patients with hyperlipoproteinaemia (Manninen 1983). Of the 25 531 men invited to participate in the trial, 18966 (80.6%) came to the first screening. A complete lipid profile for this cohort is available and has been partly presented elsewhere (Manninen et al. 1989). 6903 men met the lipid acceptance criterion of a non-HDL-cholesterol level (or atherogenic cholesterol) exceeding 5.2 mmol/L. (NonHDL-cholesterol = total cholesterol minus HDLcholesterol.) Subjects accepted into the study were required to meet this criterion on 2 successive determinations. This led to a final study group of 4081 subjects who were randomly allocated to receive
,,, ,
35
A
30 25
;g
20
\
\
0
;; 15
\
\
\
\
U
.'11. 10
.c
~
C/l
\
"-
"-
5 0
2
3
4
5
6
7
8
\
\
9
\
10
Total cholesterol (mmol/L) Fig. 1. Distribution of total cholesterol of 18966 screened men (-) and 4081 trial participants (-----) [from Manninen et al. 1989, with permission].
either gemfibrozil (2046 men) or a matching placebo (2035 men). This paper is based on the analyses of the risk-factor function of HDL-cholesterol in the placebo group. The methods used to determine plasma lipid levels, and the assumptions in calculations used to derive LDL-cholesterol have been published previously (Manninen et al. 1988). Data on HDL- and LDL-cholesterol were categorised by use of tertile limits for the whole trial population as cut-off points (table I). End-points used in the study were non35 30 25
Table I. Categorisation of the Helsinki Heart Study population based on plasma HDL- and LDL-cholesterol levels Tertile
Plasma lipoprotein level (mmol/L)
20
l15 II>
~ 10
E
~
II III
HDL-cholesterol
LDL-cholesterol
< 1.08
< 4.5
1.08-1.32 > 1.32
4.5-5.2 > 5.2
Abbreviations: HDL = high density lipoprotein; LDL = low density
lipoprotein.
C/l
5 0
0
0.4
0.8
1.2
1.6
2.0
2.4
HDL-cholesterol (mmol/L) Fig. 2. Distribution of high density lipoprotein (HDL)-cholesterol of 18966 screened men (-) and 4081 trial participants (---) [from Manninen et al. 1989, with permission].
HDL-Cholesterol as a Risk Factor in Coronary Heart Disease
2.8
2.8
ec
ttl
1.6
~ ~
:g
~ 1.2
.c:
" Cl
" C1l .J>-
~ ~
2.0
E
g "2 ~ C1l a'" .J
0.8
:I:
~'I 0.8
1.2
I
1.6
i
2.0
i
2.4
I
2.8
HDL cholesterol (mmol/L) 1st year mean
b
2.4
2: a
2.0
E
g C1l
a
2.4
2: a
9
c: 1.6 ttl C1l
E ~
C1l >-
1.2
.c:
.r; 0.8 0.8
I
1.2
1.6
i
2.0
i
2.4
I
2.8
HDL cholesterol (mmol/L) 1st year mean
Fig. 3. Mean yearly high density lipoprotein (HDL)-cholesterol at (a) year I vs year 2. and (b) year I vs year 5 during the follow-up of the placebo group of the Helsinki Heart Study (from Manninen et al. 1989. with permission).
fatal or fatal myocardial infarction, and sudden cardiac death. The effects of lipoprotein levels on the incidence of coronary heart disease (number of cardiac end-points per 1000 person-years) were analysed. Risk patterns of the joint effect of lipoproteins and other coronary risk factors were also studied using Cox Proportional Hazard models (Kalbfleisch & Prentice 1980).
2. Results The lipid distribution curves for total cholesterol of the screened population of 18 966 men and of the trial participants (n = 4081) revealed that the latter represented a high total cholesterol population (fig. 1). LDL-cholesterol levels were also higher in trial participants. However, similar curves for HDL-cholesterol showed that the distribution of HDL-cholesterol in the trial participants was similar to that in the population at large (fig. 2) suggesting that HDL-cholesterol distribution is largely independent of total cholesterol distribution. When the seasonal variation in HDL was eliminated by calculating yearly mean values, it became evident that the HDL remained very stable throughout the 5-year follow-up in the placebo group (fig. 3).
When HLD-cholesterol tertiles were plotted against LDL-cholesterol tertiles it was evident that both lipid components were related to the occurrence of coronary heart disease (fig. 4). The cardiac end-point rate was low (5.4 per 1000 person-years) in the highest HDL-cholesterol tertile (> 1.32 mmoljL) and increased to 7.2 per 1000 person-years as HDL-cholesterol levels were reduced « 1.08 mmol/L). Within the lowest HDLcholesterol tertile, the incidence of coronary heart disease increased as LDL-cholesterol levels increased. When the effect of HDL on coronary heart disease incidence was studied jointly with age, smoking and blood pressure, a similar inverse relationship was observed (fig. 5). Patients aged> 47 years had a higher rate of coronary heart disease than younger patients « 47 years) at any level ofHDLcholesterol. However, in both age groups, the number of cardiac end-points decreased as HDL-cholesterol levels increased. Similarly, in patients with blood pressure > l30/90mm Hg, the risk of coronary heart disease was reduced in patients with high HDL-cholesterol levels. Finally, in smokers, high HDL-cholesterollevels protected against coronary heart disease end-points, the incidence rate in smokers with high HDL-cholesterol being sim-
Drugs 40 (Suppl. 1) 1990
10
20
a
15 10 5
0
,;; 47 years Age
2
> 47 years
3
HDL-cholesterol terti Ie Fig_ 4_ Crude incidences per 1000 person-years of coronary heart disease in the placebo group of the Helsinki Heart Study (from Manninen et al. 1989, with permission). LDL = low density lipoprotein; HDL = high density lipoprotein.
Q)
V>
'"
20
b
Q)
V>
U
1:
'" Q)
15
~
~
c: e'"0
10
0
ilar to that in non-smokers at low or medium HDLcholesterol levels. A somewhat weaker relationship was observed between LDL-cholesterol and coronary heart disease risk (fig_ 6). When triglycerides were added to the HDL-cholesterol model, the risk pattern remained more or less unchanged.
3. Discussion Multiple blood sampling over a 5-year period of the Helsinki Heart Study enabled the long term stability ofHDL-cholesterollevels to be examined. The use of yearly mean values to eliminate seasonal variation resulted in a remarkable stability of HDL-cholesterol levels over time. The Framingham Heart Study also found a good stability of HDL-cholesterol over 8 years of follow-up (Abbott et al. 1988). The data on HDL-cholesterol and coronary heart disease risk showed that HDL-cholesterol isa strong predictorofthis disease. High HD Lcholesterol levels were clearly protective while low HDL-cholesterollevels markedly increased the rate of coronary heart disease. When analysed jointly with age, smoking and blood pressure, this inverse
15 Q)
0
c:
5
Q)
'C
'0
E
0
20
Non/past smokers
Current smokers
c
< 130/90mm Hg
> 130/90mm Hg Blood pressure
Fig. 5. Incidence of coronary heart disease (number of cardiac end-points per 1000 person-years) in baseline high density lipoprotein-cholesterol tertiles (0 = < 1.08 mmolfL, ~ = 1.08-1.32 mmol/L,. = > 1.32 mmolfL) analysed jointly with (a) age, (b) smoking habits, or (c) blood pressure (from Manninen et al. 1989, with permission).
HDL-Cholesterol as a Risk Factor in Coronary Heart Disease
20
a
15
10
5
0
> 47 years
,;; 47 years Age
'"'"
20
t::
15
Q) Q)
'" '" s::;
'5 Q)
b
~
c e'0"
10
'0 Q)
c
5
Q)
'0 '(3
E
0 Non/past smokers
20
Current smokers
c
11
relationship persisted, with no significant change seen when triglycerides were added to the model. The less clearcut association ofLDL-cholesterol and coronary heart disease risk is presumably due to the fact that low LDL-cholesterol levels were excluded by the patient selection procedure (only patients with non-HDL-cholesterol levels > 5.2 mmol/L were included), and the risk profile could therefore not be analysed over the entire LDL distribution curve. On the other hand, HDLcholesterol distribution in the study group represented the population at large and was more suited for analyses over the entire range of HDLcholesterol data. In previous studies, smokers have had lower HDL-cholesterol levels than nonsmokers (Heiss et al. 1980; Robinson et al. 1987). This finding was confirmed in the present study. A particularly dangerous combination seemed to be low HDLcholesterol levels in current smokers. High HDL-cholesterol levels and smoking did not present any significant risk: the risk in this group was the same as that in nonsmokers in low and middle tertiles of HDL-cholesterol distribution. In conclusion, the present data on HDLcholesterol and coronary heart disease risk provide strong evidence for the protective role of HDLcholesterol (or some of its subfractions) in the development of coronary heart disease.
References
15
Abbott RD, Wilson PWF, Kannel WB, Castelli WP. High density lipoprotein cholesterol, total cholesterol screening, and myocardial infarction. The Framingham Study. Arteriosclerosis 8: 207-211, 1988
< 130/90mm Hg
> 130/90mm Hg
Blood pressure Fig. 6. Incidence of coronary heart disease (number of cardiac end-points per 1000 person-years) in baseline low density lipoprotein-cholesterol tertiles (0 = < 4.5 mmolfL, !21 = 4.5-5.2 mmol/L, • = > 5.2 mmol/L) analysed jointly with (a) age, (b) smoking habits, or (c) blood pressure (from Manninen et al. 1989, with permission).
Barr DP, Russ EM, Eder HA. Protein-lipid relationship in human plasma. II. In atherosclerosis and related conditions. American Journal of Medicine II: 480, 1951 Berg K, B0rresen A-L, Frick MH, Dahlen G. Serum-HDL in atherosclerotic heart-disease. Lancet 1: 10 14, 1976 Frick MH, Elo 0, Haapa K, Heinonen OP, Heinsalmi P, et al. Helsinki Heart Study: primary-prevention trial with gemfibrozil in middle-aged men with dyslipidemia. Safety of treatment, changes in risk factors, and incidence of coronary heart disease. New England Journal of Medicine 317: 1237-1245, 1987 Heiss G, Johnson NJ, Reiland S, Davis CE, Tyroler HA. The epidemiology of plasma high-density lipoprotein cholesterol levels. The Lipid Research Clinics Program Prevalence Study. Circulation 62 (Suppl. IV): 116-136,1980 Kalbfleisch JD, Prentice RL. The statistical analysis offailure time data, John Wiley and Sons Inc., New York, 1980 Manninen V. Clinical results with gemfibrozil and background
Drugs 40 (Suppl. 1) 1990
12
for the Helsinki Heart Study. American Journal of Cardiology 52: 35B-38B, 1983 Manninen V, Elo 0, Frick MH, Haapa K, Heinonen OP, et a!. Lipid alterations and decline in the incidence of coronary heart disease in the Helsinki Heart Study. Journal of the American Medical Association 260: 641-651, 1988 Manninen V, Huttunen JK, Tenkanen L, et a!. High-density lipoprotein cholesterol as a risk factor for coronary heart disease in the Helsinki Heart Study. In High density lipoproteins and atherosclerosis II, Excerpta Medica pp. 35-42, Elsevier Science Publishers, Amsterdam, 1989 Miinttiiri M, Elo 0, Frick MH, Haapa K, Heinonen OP, et a!. The Helsinki Heart Study: basic design and randomization procedure. European Heart Journal 8 (Supp!. I): 1-29, 1987 Martin MJ, Hulley SB, Browner WS, Kuller LH, Wentworth D. Serum cholesterol, blood pressure, and mortality: implications from a cohort of 361,662 men. Lancet 2: 933-936, 1986 Miller GJ, Miller NE. Plasma-high-density lipoprotein concentration and development of ischaemic heart-disease. Lancet I: 16-19, 1975 Nikkilii E. Studies on the lipid-protein relationship in normal and pathological sera and the effect of heparin on serum lipoproteins. Scandinavian Journal of Clinical and Laboratory Investigation 5 (Supp!. 8): I, 1953 Olsson AG, Rossner S, Walldius G, Carlson LA. Effect of gemfibrozil on lipoprotein concentrations in different types of hyperlipoproteinemia. Proceedings of the Royal Society of Medicine 69 (Supp!. 2): 28-31, 1976 Pearson TA, Bulkley BH, AchuffSC, Kwiterovich PO, Gordis L. The association of low levels of HDL cholesterol and arteriographically defined coronary artery disease. American Journal of Epidemiology 109: 285-295, 1979 Robinson D, Ferns GAA, Bevan EA, Stocks J, Williams PT, et a!. High density lipoprotein subtractions and coronary risk factors in normal men. Arteriosclerosis 7: 341-346, 1987 Ross R. The pathogenesis of atherosclerosis - an update. New England Journal of Medicine 314: 488-500, 1986
Correspondence and reprints: Prof. M.H. Frick, First Department of Medicine, University Central Hospital, Helsinki, Finland.
Discussion Prof. G. Assmann (Federal Republic of Germany): The relationship between HDL-cholesterol and coronary heart disease is emphasised in the Helsinki Heart Study. There is evidence from other studies that remnants of triglyceride particles have a role in the pathophysiology of coronary heart disease. Is it possible that the effect of triglycerides was overlooked in the Helsinki Heart Study because of the statistical methods used? Prof. M.H. Frick (Finland): We did Cox Proportional Hazard analyses to identify relationships between risk factors and reduction in cardiac endpoints in 26 different ways. Triglycerides were not a significant variable. We don't know why. However, the greatest reduction in cardiac end-points was seen in the group of patients with high levels of cholesterol and triglycerides (Fredrickson hyperlipoproteinaemia type lIB), suggesting that triglycerides may have a significant role. We are currently rethinking ways to analyse the data in order to evaluate this further. Dr B. Isle (Scotland): Do your results show a benefit for gemfibrozil in all-cause mortality at 6.5 years? Prof. Frick: No, all-cause mortality was not changed after 5 years or 6.5 years of gemfibrozil therapy.
HDL-Cholesterol as a Risk Factor in Coronary Heart Disease An Update of the Helsinki Heart Study
M.H. Frick, V. Manninen, 1.K. Huttunen, OP. Heinonen,
L. Tenkanen and M. Miinttiiri
First Department of Medicine, University Central Hospital and National Public Health Institute, Helsinki, Finland
Summary
The aim of the Helsinki Heart Study, a 5-year primary prevention placebo-controlled study involving 4081 dyslipidaemic men (aged 40 to 55 years), was to investigate if increasing high density lipoprotein (HDL)-cholesterol plasma levels and decreasing low density lipoprotein (LDL)-cholesterollevels would reduce the incidence of coronary heart disease. Gemfibrozil 600mg twice daily was administered to induce these changes in lipoprotein levels. Baseline HDL-cholesterollevels in the study group were similar to those in the general population. Data from patients treated with placebo were analysed to investigate the influence of HDL-cholesterol levels on the incidence of coronary heart disease. Using the number of cardiac end-points per 1000 person-years to indicate the risk of coronary heart disease, it was clear that elevated HDL-cholesterol levels reduced the risk of coronary heart disease while the incidence increased at low HDL-cholesterol levels. This relationship was not altered when the effect of HDL-cholesterollevels was analysed jointly with other coronary risk factors (age, smoking or blood pressure). A weaker association was seen between LDL-cholesterol and risk of coronary heart disease, and triglycerides appeared to have no significant effect on the incidence of the disease. The data clearly suggest that HDL-cholesterol is a strong predictor of the incidence of coronary heart disease in the placebo group of the Helsinki Heart Study.
Evidence from large scale epidemiological surveys and controlled clinical trials positively links low density lipoprotein (LDL)-cholesterol to the atherogenic process in different vascular regions, notably the coronary arteries (Martin et al. 1986; Ross 1986). The primary evidence for the protective effect of high density lipoprotein (HDL)-cholesterol was found some 35 years ago (Barr et al. 1951; Nikkilii 1953) and has been strengthened by
several epidemiological studies (Abbot et al. 1988; Miller & Miller 1975). Evidence from studies using coronary angiography is also in line with this reasoning (Berg et al. 1975; Pearson et al. 1979). This paper analyses the impact of HDL-cholesterol on the incidence of coronary heart disease in a 5-year follow-up study of dyslipidaemic, middle-aged males in the Helsinki Heart Study (Frick et al. 1987; Manninen et al. 1988; Miinttiiri et al. .1987).
8
Drugs 40 (Suppi. 1) 1990
1. Materials and Methods
40
The Helsinki Heart Study was conducted as a double-blind, placebo-controlled primary prevention study to assess the effects of gemfibrozil 600mg administered twice daily in middle-aged (40 to 55 years) dyslipidaemic men. The design and primary results have been published previously (Frick et al. 1987; Manninen et al. 1988; Manttari et al. 1987). Briefly, the basic aim of the study was to test the hypothesis that the incidence of coronary heart disease can be reduced by increasing the plasma level of HDL-cholesterol, while simultaneously decreasing LDL-cholesterol levels. Gemfibrozil was chosen for this long term (5-year) study since it is well tolerated after continued use (Olsson et al. 1976) and is an effective lipid modifier. The drug increases HDL-cholesterol and decreases total and LDL-cholesterol levels in healthy volunteers and in patients with hyperlipoproteinaemia (Manninen 1983). Of the 25 531 men invited to participate in the trial, 18966 (80.6%) came to the first screening. A complete lipid profile for this cohort is available and has been partly presented elsewhere (Manninen et al. 1989). 6903 men met the lipid acceptance criterion of a non-HDL-cholesterol level (or atherogenic cholesterol) exceeding 5.2 mmol/L. (NonHDL-cholesterol = total cholesterol minus HDLcholesterol.) Subjects accepted into the study were required to meet this criterion on 2 successive determinations. This led to a final study group of 4081 subjects who were randomly allocated to receive
,,, ,
35
A
30 25
;g
20
\
\
0
;; 15
\
\
\
\
U
.'11. 10
.c
~
C/l
\
"-
"-
5 0
2
3
4
5
6
7
8
\
\
9
\
10
Total cholesterol (mmol/L) Fig. 1. Distribution of total cholesterol of 18966 screened men (-) and 4081 trial participants (-----) [from Manninen et al. 1989, with permission].
either gemfibrozil (2046 men) or a matching placebo (2035 men). This paper is based on the analyses of the risk-factor function of HDL-cholesterol in the placebo group. The methods used to determine plasma lipid levels, and the assumptions in calculations used to derive LDL-cholesterol have been published previously (Manninen et al. 1988). Data on HDL- and LDL-cholesterol were categorised by use of tertile limits for the whole trial population as cut-off points (table I). End-points used in the study were non35 30 25
Table I. Categorisation of the Helsinki Heart Study population based on plasma HDL- and LDL-cholesterol levels Tertile
Plasma lipoprotein level (mmol/L)
20
l15 II>
~ 10
E
~
II III
HDL-cholesterol
LDL-cholesterol
< 1.08
< 4.5
1.08-1.32 > 1.32
4.5-5.2 > 5.2
Abbreviations: HDL = high density lipoprotein; LDL = low density
lipoprotein.
C/l
5 0
0
0.4
0.8
1.2
1.6
2.0
2.4
HDL-cholesterol (mmol/L) Fig. 2. Distribution of high density lipoprotein (HDL)-cholesterol of 18966 screened men (-) and 4081 trial participants (---) [from Manninen et al. 1989, with permission].
HDL-Cholesterol as a Risk Factor in Coronary Heart Disease
2.8
2.8
ec
ttl
1.6
~ ~
:g
~ 1.2
.c:
" Cl
" C1l .J>-
~ ~
2.0
E
g "2 ~ C1l a'" .J
0.8
:I:
~'I 0.8
1.2
I
1.6
i
2.0
i
2.4
I
2.8
HDL cholesterol (mmol/L) 1st year mean
b
2.4
2: a
2.0
E
g C1l
a
2.4
2: a
9
c: 1.6 ttl C1l
E ~
C1l >-
1.2
.c:
.r; 0.8 0.8
I
1.2
1.6
i
2.0
i
2.4
I
2.8
HDL cholesterol (mmol/L) 1st year mean
Fig. 3. Mean yearly high density lipoprotein (HDL)-cholesterol at (a) year I vs year 2. and (b) year I vs year 5 during the follow-up of the placebo group of the Helsinki Heart Study (from Manninen et al. 1989. with permission).
fatal or fatal myocardial infarction, and sudden cardiac death. The effects of lipoprotein levels on the incidence of coronary heart disease (number of cardiac end-points per 1000 person-years) were analysed. Risk patterns of the joint effect of lipoproteins and other coronary risk factors were also studied using Cox Proportional Hazard models (Kalbfleisch & Prentice 1980).
2. Results The lipid distribution curves for total cholesterol of the screened population of 18 966 men and of the trial participants (n = 4081) revealed that the latter represented a high total cholesterol population (fig. 1). LDL-cholesterol levels were also higher in trial participants. However, similar curves for HDL-cholesterol showed that the distribution of HDL-cholesterol in the trial participants was similar to that in the population at large (fig. 2) suggesting that HDL-cholesterol distribution is largely independent of total cholesterol distribution. When the seasonal variation in HDL was eliminated by calculating yearly mean values, it became evident that the HDL remained very stable throughout the 5-year follow-up in the placebo group (fig. 3).
When HLD-cholesterol tertiles were plotted against LDL-cholesterol tertiles it was evident that both lipid components were related to the occurrence of coronary heart disease (fig. 4). The cardiac end-point rate was low (5.4 per 1000 person-years) in the highest HDL-cholesterol tertile (> 1.32 mmoljL) and increased to 7.2 per 1000 person-years as HDL-cholesterol levels were reduced « 1.08 mmol/L). Within the lowest HDLcholesterol tertile, the incidence of coronary heart disease increased as LDL-cholesterol levels increased. When the effect of HDL on coronary heart disease incidence was studied jointly with age, smoking and blood pressure, a similar inverse relationship was observed (fig. 5). Patients aged> 47 years had a higher rate of coronary heart disease than younger patients « 47 years) at any level ofHDLcholesterol. However, in both age groups, the number of cardiac end-points decreased as HDL-cholesterol levels increased. Similarly, in patients with blood pressure > l30/90mm Hg, the risk of coronary heart disease was reduced in patients with high HDL-cholesterol levels. Finally, in smokers, high HDL-cholesterollevels protected against coronary heart disease end-points, the incidence rate in smokers with high HDL-cholesterol being sim-
Drugs 40 (Suppl. 1) 1990
10
20
a
15 10 5
0
,;; 47 years Age
2
> 47 years
3
HDL-cholesterol terti Ie Fig_ 4_ Crude incidences per 1000 person-years of coronary heart disease in the placebo group of the Helsinki Heart Study (from Manninen et al. 1989, with permission). LDL = low density lipoprotein; HDL = high density lipoprotein.
Q)
V>
'"
20
b
Q)
V>
U
1:
'" Q)
15
~
~
c: e'"0
10
0
ilar to that in non-smokers at low or medium HDLcholesterol levels. A somewhat weaker relationship was observed between LDL-cholesterol and coronary heart disease risk (fig_ 6). When triglycerides were added to the HDL-cholesterol model, the risk pattern remained more or less unchanged.
3. Discussion Multiple blood sampling over a 5-year period of the Helsinki Heart Study enabled the long term stability ofHDL-cholesterollevels to be examined. The use of yearly mean values to eliminate seasonal variation resulted in a remarkable stability of HDL-cholesterol levels over time. The Framingham Heart Study also found a good stability of HDL-cholesterol over 8 years of follow-up (Abbott et al. 1988). The data on HDL-cholesterol and coronary heart disease risk showed that HDL-cholesterol isa strong predictorofthis disease. High HD Lcholesterol levels were clearly protective while low HDL-cholesterollevels markedly increased the rate of coronary heart disease. When analysed jointly with age, smoking and blood pressure, this inverse
15 Q)
0
c:
5
Q)
'C
'0
E
0
20
Non/past smokers
Current smokers
c
< 130/90mm Hg
> 130/90mm Hg Blood pressure
Fig. 5. Incidence of coronary heart disease (number of cardiac end-points per 1000 person-years) in baseline high density lipoprotein-cholesterol tertiles (0 = < 1.08 mmolfL, ~ = 1.08-1.32 mmol/L,. = > 1.32 mmolfL) analysed jointly with (a) age, (b) smoking habits, or (c) blood pressure (from Manninen et al. 1989, with permission).
HDL-Cholesterol as a Risk Factor in Coronary Heart Disease
20
a
15
10
5
0
> 47 years
,;; 47 years Age
'"'"
20
t::
15
Q) Q)
'" '" s::;
'5 Q)
b
~
c e'0"
10
'0 Q)
c
5
Q)
'0 '(3
E
0 Non/past smokers
20
Current smokers
c
11
relationship persisted, with no significant change seen when triglycerides were added to the model. The less clearcut association ofLDL-cholesterol and coronary heart disease risk is presumably due to the fact that low LDL-cholesterol levels were excluded by the patient selection procedure (only patients with non-HDL-cholesterol levels > 5.2 mmol/L were included), and the risk profile could therefore not be analysed over the entire LDL distribution curve. On the other hand, HDLcholesterol distribution in the study group represented the population at large and was more suited for analyses over the entire range of HDLcholesterol data. In previous studies, smokers have had lower HDL-cholesterol levels than nonsmokers (Heiss et al. 1980; Robinson et al. 1987). This finding was confirmed in the present study. A particularly dangerous combination seemed to be low HDLcholesterol levels in current smokers. High HDL-cholesterol levels and smoking did not present any significant risk: the risk in this group was the same as that in nonsmokers in low and middle tertiles of HDL-cholesterol distribution. In conclusion, the present data on HDLcholesterol and coronary heart disease risk provide strong evidence for the protective role of HDLcholesterol (or some of its subfractions) in the development of coronary heart disease.
References
15
Abbott RD, Wilson PWF, Kannel WB, Castelli WP. High density lipoprotein cholesterol, total cholesterol screening, and myocardial infarction. The Framingham Study. Arteriosclerosis 8: 207-211, 1988
< 130/90mm Hg
> 130/90mm Hg
Blood pressure Fig. 6. Incidence of coronary heart disease (number of cardiac end-points per 1000 person-years) in baseline low density lipoprotein-cholesterol tertiles (0 = < 4.5 mmolfL, !21 = 4.5-5.2 mmol/L, • = > 5.2 mmol/L) analysed jointly with (a) age, (b) smoking habits, or (c) blood pressure (from Manninen et al. 1989, with permission).
Barr DP, Russ EM, Eder HA. Protein-lipid relationship in human plasma. II. In atherosclerosis and related conditions. American Journal of Medicine II: 480, 1951 Berg K, B0rresen A-L, Frick MH, Dahlen G. Serum-HDL in atherosclerotic heart-disease. Lancet 1: 10 14, 1976 Frick MH, Elo 0, Haapa K, Heinonen OP, Heinsalmi P, et al. Helsinki Heart Study: primary-prevention trial with gemfibrozil in middle-aged men with dyslipidemia. Safety of treatment, changes in risk factors, and incidence of coronary heart disease. New England Journal of Medicine 317: 1237-1245, 1987 Heiss G, Johnson NJ, Reiland S, Davis CE, Tyroler HA. The epidemiology of plasma high-density lipoprotein cholesterol levels. The Lipid Research Clinics Program Prevalence Study. Circulation 62 (Suppl. IV): 116-136,1980 Kalbfleisch JD, Prentice RL. The statistical analysis offailure time data, John Wiley and Sons Inc., New York, 1980 Manninen V. Clinical results with gemfibrozil and background
Drugs 40 (Suppl. 1) 1990
12
for the Helsinki Heart Study. American Journal of Cardiology 52: 35B-38B, 1983 Manninen V, Elo 0, Frick MH, Haapa K, Heinonen OP, et a!. Lipid alterations and decline in the incidence of coronary heart disease in the Helsinki Heart Study. Journal of the American Medical Association 260: 641-651, 1988 Manninen V, Huttunen JK, Tenkanen L, et a!. High-density lipoprotein cholesterol as a risk factor for coronary heart disease in the Helsinki Heart Study. In High density lipoproteins and atherosclerosis II, Excerpta Medica pp. 35-42, Elsevier Science Publishers, Amsterdam, 1989 Miinttiiri M, Elo 0, Frick MH, Haapa K, Heinonen OP, et a!. The Helsinki Heart Study: basic design and randomization procedure. European Heart Journal 8 (Supp!. I): 1-29, 1987 Martin MJ, Hulley SB, Browner WS, Kuller LH, Wentworth D. Serum cholesterol, blood pressure, and mortality: implications from a cohort of 361,662 men. Lancet 2: 933-936, 1986 Miller GJ, Miller NE. Plasma-high-density lipoprotein concentration and development of ischaemic heart-disease. Lancet I: 16-19, 1975 Nikkilii E. Studies on the lipid-protein relationship in normal and pathological sera and the effect of heparin on serum lipoproteins. Scandinavian Journal of Clinical and Laboratory Investigation 5 (Supp!. 8): I, 1953 Olsson AG, Rossner S, Walldius G, Carlson LA. Effect of gemfibrozil on lipoprotein concentrations in different types of hyperlipoproteinemia. Proceedings of the Royal Society of Medicine 69 (Supp!. 2): 28-31, 1976 Pearson TA, Bulkley BH, AchuffSC, Kwiterovich PO, Gordis L. The association of low levels of HDL cholesterol and arteriographically defined coronary artery disease. American Journal of Epidemiology 109: 285-295, 1979 Robinson D, Ferns GAA, Bevan EA, Stocks J, Williams PT, et a!. High density lipoprotein subtractions and coronary risk factors in normal men. Arteriosclerosis 7: 341-346, 1987 Ross R. The pathogenesis of atherosclerosis - an update. New England Journal of Medicine 314: 488-500, 1986
Correspondence and reprints: Prof. M.H. Frick, First Department of Medicine, University Central Hospital, Helsinki, Finland.
Discussion Prof. G. Assmann (Federal Republic of Germany): The relationship between HDL-cholesterol and coronary heart disease is emphasised in the Helsinki Heart Study. There is evidence from other studies that remnants of triglyceride particles have a role in the pathophysiology of coronary heart disease. Is it possible that the effect of triglycerides was overlooked in the Helsinki Heart Study because of the statistical methods used? Prof. M.H. Frick (Finland): We did Cox Proportional Hazard analyses to identify relationships between risk factors and reduction in cardiac endpoints in 26 different ways. Triglycerides were not a significant variable. We don't know why. However, the greatest reduction in cardiac end-points was seen in the group of patients with high levels of cholesterol and triglycerides (Fredrickson hyperlipoproteinaemia type lIB), suggesting that triglycerides may have a significant role. We are currently rethinking ways to analyse the data in order to evaluate this further. Dr B. Isle (Scotland): Do your results show a benefit for gemfibrozil in all-cause mortality at 6.5 years? Prof. Frick: No, all-cause mortality was not changed after 5 years or 6.5 years of gemfibrozil therapy.
