Lipids and lipoproteins as predictors of coronary heart disease, stroke, and cancer in the honolulu...

Lipids and Lipoproteins as Predictors of Coronary Heart Disease, Stroke, and Cancer in the Honolulu Heart Program

DWAYNE REED, M.D., Ph.D. KATSUHIKO YANO, M.D. ABRAHAM KAGAN, M.D. Honolulu, Hawaii

From the National Heart, Lung, and Blood Insti- tute, Honolulu Heart Program, and the Kuakini Medical Center, Honolulu Heart Program, Hono- lulu, Hawaii. This work was supported by Con- tract NOI-HV-2901 from the National Heart, Lung, and Blood Institute. Requests for reprints should be addressed to Dr. Dwayne Reed, Hono- lulu Heart Program, 347 North Kuakini Street, Honolulu, Hawaii 96817. Manuscript accepted May 2, 1985.

A group of 2,122 healthy men in the Honolulu Heart Program who participated in the Cooperative Lipoprotein Phenotyping Study, 1970 to 1972, were followed for 10 years by repeated examinations and surveillance of hospital discharge and mortality records in order to diagnose new cases of coronary heart disease, stroke, cancer, and other deaths. Total cholesterol and low-density lipoprotein cholesterol were significantly associated with all clinical types of coronary heart disease in multivariate analyses, whereas high-density lipoprotein cholesterol was inversely associated with nonfatal myocardial infarction and total coronary heart disease, but not with fatal coronary heart disease nor angina. Triglyceride and very-low-density lipoprotein cholesterol were associated with total coronary heart disease by univariate but not multivariate analysis. None of the other specific chronic diseases were significantly associated with any lipid or lipoprotein, although there were trends of inverse associations of all noncardiovascular disease with total cholesterol and low-density lipoprotein cholesterol. Thus, for total disease (coronary heart disease, stroke, cancer, and other deaths), the optimal range for lowest disease incidence was about 200 to 220 mg/dl for total cholesterol and 120 to 140 mg/dl for low-density lipoprotein cholesterol. A strong inverse pattern of total disease with high-density lipoprotein cholesterol indicated that the highest levels were the optimal levels.

In 1970, a sample of men in the Honolulu Heart Program participated in the Cooperative Lipoprotein Phenotyping Study [ 1,2]. Since that time, these men have been followed by two repeated examinations five and 10 years after the phenotyping study, and by continuous surveillance of hospital discharge and mortality records. It was thus possible to identify all new cases of coronary heart disease, stroke, cancer, and deaths due to all other causes.

This report presents analyses of the association of the lo-year incidence of coronary heart disease with the baseline lipid and lipoprotein levels. As consideration of such data may miss the overall relationship of lipoproteins to chronic disease in general, and cancer in particular, we have also included data on the 1 O-year incidence of stroke, cancer, and all other mortality.

SUBJECTS AND METHODS

Study Population. The Honolulu Heart Program is a long-term prospective epidemiologic investigation of coronary heart disease and stroke among men of Japanese ancestry who were born during 1900 to 19 19 and were living on Oahu Island, Hawaii, in 1965. During 1965 to 1968, 8,006 eligible men participated in the examination. Details of the study proce-

May 1986 The American Journal of Medicine Volume 80 871

LIPOPROTEINS AND CORONARY HEART DISEASE-REED ET AL

dures and the cohort characteristics have been described elsewhere [3,4]. Approximately 94 percent of these men were re-examined two years after their initial examination. Towards the end of this second-cycle examination in 1970, we were invited to take part in the Cooperative Lipoprotein Phenotyping Study in which levels of blood lipids and lipoproteins were compared among six population-based epidemiologic studies in the United States using a common protocol for determinations of lipids and lipoproteins [I].

In the Honolulu Heart Program, three separate groups of men who participated in the second-cycle examination (1968 to 1970) were selected for this collaborative study: (1) a 30 percent probability sample (n = 1,859), (2) men who had serum cholesterol or triglyceride levels in the highest 10 percent at the initial examination (n = 707), and (3) prevalence cases of coronary heart disease or stroke ascertained at the second-cycle examination (n = 214), for a total of 2,780 men, aged 50 to 7 1.

The lipoprotein examinations were performed between 1970 and 1972. Survivors were re-examined from 1975 to 1978 and again from 1980 to 1982. For this lo-year incidence study, all men with prevalent coronary heart disease, stroke, or cancer at baseline were excluded, as were men who were being treated for hypercholesterolemia or diabe- tes. Thus, 658 men were dropped from the study, leaving 2,122, including 1,545 men in the probability sample, 344 men with high cholesterol levels, and 233 with high triglyceride levels. Men with both high cholesterol and triglyceride levels were counted only in the high cholesterol group.

At the baseline lipoprotein examination (1970 to 1972), information was obtained only on medical history, drug use, special diets, alcohol consumption, and weight changes. Data for other coronary heart disease risk factors were obtained from the second-cycle examination (1968 to 1970), except for serum glucose and uric acid, which were measured only at the initial examination (1965 to 1968). Laboratory Methods. After an overnight fast of at least 12 hours, 20 ml blood specimens were drawn into EDTA tubes. Plasma was separated by spinning in a refrigerated centri- fuge at 2,000 revolutions per minute for 15 minutes. An aliquot of plasma was refrigerated overnight and examined for surface chylomicrons as a check for fasting. The plasma samples were sent on ice to the Public Health Service Laboratory in San Francisco. The laboratory methods followed a common protocol specified by the Laboratory of Molecular Disease, National Heart, Lung, and Blood Insti- tute, as described earlier [l]. In brief, the plasma was divided into four portions: 0.5 ml for total cholesterol deter- mination, 0.5 ml for triglyceride, 3 ml for high-density lipoprotein cholesterol, and 5 ml for beta quantification.

The 5 ml used for beta quantification was drawn into a Beckman 302232 cellulose nitrate tube and carefully over- laid with 0.15 M saline. Cellulose nitrate tubes so prepared were then spun in a precooled rotor for 16 to 18 hours at 40,000 revolutions per minute in a Beckman preparative ultracentrifuge.

The tubes were then sliced, with the tube placed in the slicer so that the bottom edge of the cap was 0.8 cm from the top of the slicer. The top and bottom fractions were

transferred to 5 ml volumetric flasks and brought up to volume with 0.15 Msaline. The bottom-fraction cholesterol was determined by measuring the amount of cholesterol in an aliquot from the bottom fraction. High-density lipoprotein cholesterol was determined after precipitating low-density lipoprotein cholesterol and very-low-density lipoprotein cholesterol by heparin-manganese chloride.

The level of low-density lipoprotein cholesterol was calculated as the difference between the concentration of bottom fraction and high-density lipoprotein cholesterol. Very-low-density lipoprotein cholesterol concentration was obtained by calculating the difference between total and bottom-fraction cholesterol. Lipoprotein electrophoresis was run on whole, top, and bottom fractions. Cholesterol determinations were performed using the Technicon Au- toanalyzer I N24B method, and triglycerides were determined by the semiautomated Lederer-Kessler method. Case Finding. The identification of new cases of cardiovascular and other major chronic diseases was achieved by the follow-up examinations, and by continuous surveillance of hospitalization discharge records, the Hawaii Tumor Registry, and death certificates. Diagnoses were determined by a panel of study physicians on the basis of all available information using criteria described earlier [4]. Clinical subgroups of coronary heart disease included fatal coronary heart disease, nonfatal myocardial infarction, and combined angina pectoris and coronary insufficiency. Any- one with more than one manifestation of coronary heart disease was classified by the worst manifestation. For this report, all strokes were combined as were all types of cancer. Subgroups of these diseases will be examined in later reports. Statistical Methods. The analytic procedures used to ex- amine the relationship of baseline lipoprotein levels to the lo-year disease incidence included comparisons of the baseline characteristics and disease incidence rates of the high cholesterol and high triglyceride groups with those of the random sample group, and more detailed analyses of the men in the random sample.

Age-adjusted lo-year incidence rates of coronary heart disease, stroke, cancer, and other deaths were calculated by quartiles of baseline lipoprotein levels. Potential confounding variables were selected from the risk factors known to be related to coronary heart disease in this cohort [4]. These included age, blood pressure, body mass index, forced vital capacity, physical activity index, cigarettes smoked per day, alcohol consumption per week, serum glucose, and serum uric acid. Forced vital capacity was not included as a confounder as it was not associated with the lipoproteins. Tests of age-adjusted mean values were made using analyses of covariance and Mantel-Haenszel chi- square tests.

The independent role of the lipoproteins in the prediction of coronary heart disease was evaluated using a multivariate logistic model. Parameters of the logistic function were estimated using an iterative maximum likelihood technique. The logistic coefficients for the independent variables were standardized by multiplying them by the standard deviations of the respective variables. The significance of each coeffi-

a72 May 1986 The American Journal of Medicine Volume 80

LIPOPROTEINS AND CORONARY HEART DISEASE--REED ET AL

cient was tested by dividing it by its standard error and relating this statistic (Z value) to a table of standard normal deviates using the two-tailed significance level.

RESULTS

Table I shows the age-adjusted mean levels of lipids, lipoproteins, and other selected characteristics by study group. As reported earlier [l], the lipid and lipoprotein values for the random sample men were quite similar to those of Framingham for total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol. The triglyceride levels were much higher than those for Framingham and most other centers.

The men chosen because of high cholesterol levels at the first examination had lipid levels that were significantly higher than those of the random sample for total cholesterol, low-density lipoprotein cholesterol, and triglycerides. The high triglyceride sample men differed significantly from the random sample for triglycerides and very- low-density lipoprotein cholesterol levels only. The intercorrelation of the lipoproteins and lipids for the random sample men was also nearly identical to the values published earlier [I]. Total cholesterol and low-density lipoprotein cholesterol were highly correlated (r = 0.74) as were very-low-density lipoprotein cholesterol and triglyceride (r = 0.89); high-density lipoprotein cholesterol was not associated with total cholesterol, nor low-density lipoprotein cholesterol, but it was inversely associated with both triglyceride and very-low-density lipoprotein cholesterol (r z -0.37). Triglyceride and very-low-density lipoprotein cholesterol were moderately associated with total cholesterol (~0.30) but inversely associated with low-density lipoprotein cholesterol (r g -0.30). Because of.the high correlations, total cholesterol was not included together with low-density lipoprotein cholesterol in multivariate models, nor was triglyceride included with very-low-density lipoprotein cholesterol.

The covariables shown in Table I include those that were found to be significantly related to coronary heart disease incidence in this study group. There were no significant differences between the random sample and high cholesterol sample men for any of these risk factors. The high triglyceride sample men, however, had significantly higher levels of systolic blood pressure, alcohol intake, body mass index, and serum uric acid levels.

Table II shows the age-adjusted 1 O-year disease incidence rates for the same three study groups. For all clinical subgroups of coronary heart disease, the incidence rates were significantly higher for the high cholesterol group than for the random sample men. The high triglyceride sample men had a significantly higher rate of fatal coronary heart disease only. They also had a significantly lower rate of total stroke incidence. There were no other significant differences for stroke, cancer, and other

TABLE I Age-Adjusted Mean Levels of Plasma Lipids and Lipoproteins and Other Selected Variables by Study Group

Variable

High High Random Sample Cholesterol Triglyceride

(n = 1,545) (II = 344) (n = 233)

Total cholesterol 221 268* 218 (Wdl)

High-density lipoprotein 45 46 39 cholesterol (mg/dl)

Low-density lipoprotein 143 179” 124 cholesterol (mg/dl)

Very-low-density 33 42 54” lipoprotein cholesterol (mg/dl)

Triglyceride (mg/dl) 171 223* 317” Systolic blood pressure 133 133 137*

(mm W Alcohol (ounces/week) 3.5 3.2 4.7” Body mass index+ 23.7 24.1 24.8’ Cigarettes per day? 9 10 10 Serum glucose (mg/dl)t 153 156 157 Serum uric acid (mg/dl)t 6.0 6.3 6.5” Age 58 58 58

l These values differ significantly from those of the random sample at a p value of 0.05 or less. t Data from the 1968-1970 examination. t Data from the 1965-1968 examination.

TABLE II Age-Adjusted lo-Year Incidence Rates per 1,000 Populatlon for Coronary Heart Disease and Other Chronic Diseases by Study Group

Disease

Fatal coronary heart disease

Nonfatal myocardial infarction

Angina/coronary insufficiency

Total coronary heart disease

Total stroke Total cancer Other deaths

Random Sample

20 (29)

32 (47)

37 (55)

85 (131)

29 (44) 78 (120) 29 (42)

High Cholesterol

37’ (11)

63” (20)

74” (23)

157” (54)

47 (16) 67 (23) 38 (12)

Triglyceride

45” (10)

31 (7)

37 (8)

107 (25)

4” (1) 86 (20) 29 (6)

* These values are significantly different from those of the random sample at a p value of 0.05 or less. Number of cases are shown in parentheses.

deaths. The significant differences shown in Table II per- sisted in multivariate analyses including the covariables shown in Table I.

The remaining analyses involved only the 1,545 men in the random sample. The men were first divided into quartiles of lipoprotein and lipid levels, and age-adjusted



TABLE III Age-Adjusted lo-Year Incidence Rates of Coronary Heart Disease per 1,000 Population by Quartiles of Lipids and Lipoprotein among the Random Sample

Number Fatal Nonfatal

Coronary Heart Disease hlyocardial lnfarclion Angina

Coronary lnsuff iciency Total

Coronary Heart Disease

Quartiles of Total Cholesterol (mg/dl) 103-196 395 15 197-219 386 14 220-244 379 22 245-384 385 31

p value 0.03 Quartiles of High-Density Lipoprotein Cholesterol (mg/dl)

17-37 445 20 38-43 340 17 44-53 417 29 54-103 343 12 p value NS

Quartiles of Low-Density Lipoprotein Cholesterol (mg/dl) 26-120 392 11

121-143 398 28 144-168 370 21 169-280 385 23

p value 0.03 Quartiles of Very-Low-Density Lipoprotein Cholesterol (mg/dl)

1-17 391 13 18-26 396 19 27-39 385 25 40-229 373 26 p value NS

Quartiles of Triglyceride (mg/dl) 12-93 411 19 94-131 372 la

133-194 386 17 195-1,360 376 28

p value NS

23 23 59 25 30 65 33 44 93 47 53 120

0.03 <O.OOl <o.ooi

49 53 37 27 29 31 12 34

0.005 NS

21 35 22 50

0.03

27 26 63 30 29 74 28 51 98 47 45 109 NS NS 0.04

21 la 56 28 42 a3 44 38 92 38 53 110 NS NS 0.01

21 37 45 47

0.02

113 77 a3 55

0.005

51 94 a3

111 <O.OOl

p values from logistic regression analysis with age in the model.

TABLE IV Significant Standardized Coefficients from Multiple Logistic Analyses of Coronary Heart Disease, Including a Single Lipid or Lipoprotein One at a Time with Selected Covarlables’

Fatal Nonfalal Angina Total Coronary Heart Disease Myocardial lnfarclion Coronary lnsufl iciency Coronary Heart Disease

Total cholesterol 0.360t 0.300+ 0.372$ 0.3479 High-density lipoprotein cholesterol -0.475x -0.244t Low-density lipoprotein cholesterol 0.377+ 0.326t 0.3763 0.3573 Very-low-density lipoprotein cholesterol Triglyceride Total cholesterol/high-density 0,328t 0.372$ 0.3646

lipoprotein cholesterol Low-density lipoprotein cholesterol/high- 0.356t 0.320t 0.403*

density lipoprotein cholesterol

l Covariables included age, body mass index, systolic blood pressure, serum uric acid, serum glucose, cigarettes per day, and alcohol per week. + p 10.05. t p (0.01. 5 p 10.001.

874 May 1988 The American Journal of Medicine Voluine 80


incidence rates of clinical subgroups of coronary heart disease were calculated as shown in Table III. Total cholesterol and low-density lipoprotein cholesterol were significantly associated with all clinical types of coronary heart disease, and there was a dose-response pattern for both measures, which was more consistent for total cholesterol than for low-density lipoprotein cholesterol.

High-density lipoprotein cholesterol was inversely associated with nonfatal myocardial infarction and total coronary heart disease, but not with fatal coronary heart disease nor angina and coronary insufficiency. There was a fairly consistent dose-response pattern of association for both nonfatal myocardial infarction and total coronary heart disease. The men in the lowest quartile had the highest rate of angina/coronary insufficiency, but the rates for the other quartiles were similar.

Very-low-density lipoprotein cholesterol and triglyceride were significantly associated with total coronary heart disease, but not with the clinical subgroups, although there were general patterns of increased risk with higher quartiles.

The lipoproteins were next included one at a time with selected covariables in multiple logistic analyses of the clinical groups of coronary heart disease as shown in Table IV. The significant associations were similar to those found in Table Ill, except that very-low-density lipoprotein cholesterol and triglyceride were no longer associated with total coronary heart disease. We also included the ratios of total cholesterol/high-density lipoprotein cholesterol and low-density lipoprotein cholesterol/high-density lipoprotein cholesterol in these models. Their significant associations were similar to that of the variable in the numerator, except that neither ratio was associated with fatal coronary heart disease.

High-density lipoprotein cholesterol was also included with total cholesterol or with low-density lipoprotein cholesterol in similar multiple logistic analyses of nonfatal myocardial infarction and total coronary heart disease. Both pairs of variables remained significant, with standardized coefficients nearly identical to those shown in Table IV, indicating that the inverse association of high- density lipoprotein cholesterol was independent of either total cholesterol or low-density lipoprotein cholesterol.

As total cholesterol is generally considered to be a weaker predictor of coronary heart disease than high- density lipoprotein cholesterol among older populations, we examined the IO-year incidence rates of fatal coronary heart disease and nonfatal myocardial infarction by the age groups 50 to 59, and 60 to 71. For the younger age group, the incidence rates increased significantly in a consistent manner by quartiles of total cholesterol and decreased consistently by quartiles of high-density lipoprotein cholesterol. Thus, for this study group, total cholesterol appeared to be as equally good as high-density lipoprotein cholesterol in predicting definite coronary

240

s .; 180- m

3

; 160-

0” E 140-

2 +i Lc

120-

0”

E -1 P ; 5 ,g 100 80 60 .----., \ ‘.,/ / -Lncer /*

I A” I I I 103-196 197-219 220-244 245-384

Quartiles of Total Cholesterol

Figure 1. Ten-year age-adjusted disease incidence rares by quartiles of total cholesterol. #p < 0.05 in multiple logistic regression including a quadratic expression for total cholesterol. CHD = coronary heart disease.

heart disease in the younger age group, and as equally poor in the older age group.

Figure 1 shows the age-adjusted IO-year incidence rates for stroke, cancer, other deaths, and all major chronic disease by quartiles of total cholesterol. In several cases, the distribution of rates was U-shaped, with higher rates at both the lowest and highest quartiles of total cholesterol. When tested using quadratic terms in multiple logistic models, these patterns were significant for stroke and for total disease. The patterns of distribution of disease rates by quartiles of low-density lipoprotein cholesterol were quite similar, but the association with total disease was not significant (Table V).

Figure 2 shows similar rates by quartiles of high- density lipoprotein cholesterol. There, patterns of inverse associations for cancer and stroke were not statistically significant, but they added to the significant inverse association of high-density lipoprotein cholesterol with total disease. This association remained significant in multiple logistic analyses including the covariables listed in Table IV.



TABLE V Age-Adjusted IO-Year Incidence Rates of Stroke, Cancer, and Other Deaths per 1,000 Population by Quartiles of Lipids and Lipoprotein

Total Total Other Stroke Cancer Deaths

Total Non-Coronary Heart Disease

Total * Disease

Quartiles of Total Cholesterol (mg/dl) 103-196 32 197-219 14 220-244 28 245-384 40

78 79 67 85

p value B 0.05+ NS Quartiles of High-Density Lipoprotein Cholesterol (mg/dl)

17-37 33 89 38-43 30 86 44-53 25 76 54-103 25 64 p value NS NS

Quartiles of Low-Density Lipoprotein Cholesterol (mg/dl) 26-120 37 83

121-143 15 72 144-168 22 79 169-280 41 80

p value 0.031 NS Quartiles of Very-Low-Density Lipoprotein Cholesterol (mg/dl)

1-17 27 81 18-26 26 27-39 29 40-229 34 p value NS

Quartiles of Triglyceride (mg/dl) 12-93 21 94-131 25

133-194 30 195-1,360 40

p value NS

56 115

62 NS

78 39 138 186 62 23 110 188 94 30 154 218 78 25 143 241

NS NS NS 0.03

42 152 214 22 115 173 34 129 205 20 145 233 NS NS 0.03+

32 154 255 25 141 194 21 122 191 44 133 186 NS NS 0.01

33 153 207 39 126 204 27 128 192 22 143 232 NS NS NS

43 151 203 21 163 165 29 173 254 24 120 214 NS NS NS

p values from logistic regression analysis with age in the model. * Includes both coronary heart disease and non-coronary heart disease. 7 From logistic regression including a quadratic expression. The test for linear regression was not significant.

Table V also shows the disease rates by quartiles of very-low-density lipoprotein cholesterol and triglyceride. The only significant association was for total disease with triglyceride, and this association did not remain significant in multiple logistic analyses including the covariables listed in Table IV.

COMMENTS

These analyses of the lo-year incidence of coronary heart disease and other chronic diseases in men of Japa- nese ancestry who had lipid and lipoprotein determinations from 1970 to 1972 showed several patterns that are consistent with earlier studies of cardiovascular disease in the total cohort [4-81. By multivariate analysis, total cholesterol and low-density lipoprotein cholesterol were both significantly associated with all clinical subgroups of coronary heart disease. The inclusion of other major chronic

diseases altered the pattern of association by increasing risk of total disease in the lowest quartiles of total cholesterol and low-density lipoprotein cholesterol, resulting in a “U’‘-shaped pattern of dose response. Thus, from the viewpoint of risk of total disease, the optimal levels were about 200 to 220 mg/dl for total cholesterol and 120 to 140 mg/dl for low-density lipoprotein cholesterol.

High-density lipoprotein cholesterol was inversely and independently associated with nonfatal myocardial infarction and coronary heart disease, but not with fatal coronary heart disease and angina/coronary insufficiency. Therefore, as a predictor of incident coronary heart disease in this cohort, high-density lipoprotein cholesterol was less effective than total cholesterol or low-density lipoprotein cholesterol. By age groups, high-density lipoprotein cholesterol appeared as equally good as total cholesterol in the younger group, age 50 to 59, and as equally poor in the older group, age 60 to 7 1.

876 May 1986 The American Journal of Medicine Volume 80


There were no significant associations between high- density lipoprotein cholesterol and other selected chronic diseases, but inverse trends added to the strong inverse association between high-density lipoprotein cholesterol and coronary heart disease, indicating that the optimal level was the highest level.

The findings of this study are in general accord with those reported from other prospective studies of the relationship of lipids and lipoproteins to the risk of coronary heart disease. The direct association of cholesterol and low-density lipoprotein cholesterol, and the inverse association of high-density lipoprotein cholesterol to most manifestations of coronary heart disease are well estab- lished in a variety of ethnic groups [g-16]. The limited associations of triglycerides and very-low-density lipoprotein cholesterol tend to become diluted or disappear when confounding variables or other lipids and lipoproteins are included in multivariate analyses [ 12,14-l 61.

One major difference does emerge when we compare our data with those from some other studies. Among men in Framingham observed by similar laboratory and analytic methods, high-density lipoprotein cholesterol was a more potent lipid risk factor than low-density lipoprotein cholesterol or total cholesterol [IO, 16,171. In the Israeli lschemic Heart Disease Study, high-density lipoprotein cholesterol had a stronger association with myocardial infarction among the older age group than did total cholesterol, but it was not associated with angina pectoris [ 121. Thus, despite the fact that the baseline levels of total cholesterol, low-density lipoprotein cholesterol, and high- density lipoprotein cholesterol among the men in our study were quite similar to those for the Framingham men [ 11, we did not corroborate their findings that “the protective effect of high-density lipoprotein cholesterol appears to be greater than the atherogenic effect of low-density lipoprotein cholesterol” [ 171. We have no explanation for these divergent findings.

The relationship of serum cholesterol to noncardiovascular disease and death, especially cancer, has received a great deal of attention [ 18-201. In about half of some 29 prospective studies, there was an inverse association with cancer and/or other mortality among men so that the overall association tended to be J- or U-shaped. For most of the other studies, there was no clear trend of any type. As numerous inconsistencies were noted in these patterns of associations, the general consensus of a review panel was that they do not substantiate any direct cause- and-effect relationship between low serum cholesterol levels and cancer and/or other mortality [ 191. Studies of this cohort are among those that have shown an inverse association of serum cholesterol with cancer incidence and mortality and hemorrhagic stroke [5-71. More recent- ly, these associations have been extended to include inverse associations between dietary intake of saturated

220

-T----- g ‘Z 180- m

5 ; 160-

8 0 140- : 3 %i 120- IT

240

\ xx Total Disease

Non-CHD Disease

.--- -. -.

l. l \ J

CarlCb%

‘. --.

Other deaths

I .* .’ .’

:~--~;-----.-.. .’ --__ -_ I’ --•----- -----.. --.7(..- ----. ~

1 Stroke

I I I 17 -37 38-43 44-53 54-103

Quartiles of HDL Cholesterol

Figure 2. Ten-year incidence rates of disease by quartiles of high-density lipoprotein (HDL) cholesterol. * * p < 0.0 1 from multiple logistic analyses. CHD = coronary heart disease.

fat and colon cancer [2 l] and between an index of animal food intake and stroke [22].

Other studies of lipoproteins and noncardiovascular disease are uncommon. Results from Framingham showed that low-density lipoprotein cholesterol was inversely and independently associated with noncoronary heart disease mortality in both sexes, stroke incidence in women, and cancer mortality in men [16]. High-density lipoprotein cholesterol was inversely associated with both cardiovascular and noncardiovascular disease, but these patterns were significant only for coronary heart disease.

In the Israeli lschemic Heart Disease Study, high-density lipoprotein cholesterol was inversely associated with total mortality but not with cancer in multivariate analyses [23]. Total cholesterol was inversely associated with total mortality in univariate but not multivariate analyses. In the Minnesota Prospective Study on Coronary Heart Disease, high-density lipoprotein cholesterol levels were higher among men who died of cancer than among survivors of a



25year follow-up period [24]. Competing causes of death may have been responsible for this latter finding.

The general pattern that is emerging from prospective studies of lipids and lipoproteins is that high-density lipoprotein cholesterol has a consistent inverse relationship with both cardiovascular and noncardiovascular disease, whereas total cholesterol and low-density lipoprotein cholesterol have a direct association with coronary heart

1.

5.

6.

8.

9.

10.

11.

12.

13.

disease and trends of inverse association with other chronic diseases. These trends do not satisfy our desire for consistent, independent, dose-response associations, and they conflict with other findings from ecologic and laboratory studies [20]. Yet, these trends are persistent enough to warrant our continued attention to the explora- tion of the relationship between these lipids and lipoproteins and the risk of noncardiovascular disease.

REFERENCES

Castelli C, Cooper Gr, Doyle JT: Distribution of triglyceride and total LDL-C and HDL-C in several populations: a cooperative lipoprotein phenotyping study. J Chronic Dis 1977; 30: 147.

Castelli WP, Doyle JT, Gordon T: HDL-C and other lipids in coronary heart disease. The Cooperative Lipoprotein Phenotyping Study. Circulation 1977; 55: 767.

Worth RM, Kagan A: Ascertainment of men of Japanese ancestry in Hawaii through World War II Selective Service registration. J Chronic Dis 1970; 23: 389.

Yano K, Reed DM, McGee DL: Ten-year incidence of coronary heart disease in the Honolulu Heart Program: relationship to biologic and lifestyle characteristics. Am J Epidemiol 1984; 119: 653.

Kagan A, Popper J, Rhoads GG: Factors related to stroke incidence in Hawaii Japanese men. The Honolulu Heart Study. Stroke 1980; 11: 14.

Stemmermann GN, Nomura A, Heilbrun L, Pollack ES, Ka- gan A: Serum cholesterol and colon cancer incidence in Hawaii Japanese men. JNCI 1981; 67: 1179.

Kagan, A, McGee DL,Yano K, Rhoads GG, Nomura A: Serum cholesterol and mortality in a Japanese-American population Am J Epidemiol 1981; 114: 11.

Rhoads GG, Feinleib M: Serum triglyceride and risk of coronary heart disease, stroke, and total mortality in Japa- nese-American men. Arteriosclerosis 1983; 3: 316.

Kannel WB, Castelli WP, Gordon T, McNamara P: Serum cholesterol, lipoproteins and risk of coronary heart disease: the Framingham Study. Ann intern Med 1971; 74: 1.

Gordon T, Castelli WP, Hjortland MC, Kannel WB, Dawber TR: High density lipoprotein as a protective factor against coronary heart disease: the Framingham Study. Am J Med 1977; 62: 707.

Gofman JW, Young W, Tandy R: lschemic heart disease, atherosclerosis and longevity. Circulation 1966; 34: 679.

Goldbourt U, Medalie JH: High density lipoprotein cholesterol and incidence of coronary heart disease-the Israeli lschemic Heart Disease Study. Am J Epidemiol 1979; 109: 296.

Medalie JH, Kahn HA, Neufeld HN, Riss E, Goldboult U: Five

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

year myocardial infarction incidence: II. single variables to age and birthplace. 1973; 26: 325.

Association of J Chronic Dis

Miller NE, Thelle DS, Forde OH, Mjos OD: The Tromso Heart Study: high density lipoprotein and coronary heart disease: a prospective case-control study. Lancet 1977; I: 965.

Wflhelmsen L, Wendel H, Tibblin G: Multivariate analysis of risk factors for coronary heart disease. Circulation 1973; 48: 950.

Gordon T, Kannel WB, Castelli WP, Dawber TR: Lipopro- teins, cardiovascular disease and death. The Framingham Study. Arch Intern Med 1981; 141: 1128.

Kannel W: High-density lipoproteins: epidemiologic profile and risk of coronary artery disease. Am J Cardiol 1983; 52: 9B.

Feinleib M: On a possible inverse relationship between serum cholesterol and cancer mortality. Am J Epidemiol 1981; 114: 5.

Feinleib M: Summary of a workshop on cholesterol and noncardiovascular disease mortality. Prev Med 1982; 11: 360.

Sidney S, Farquhar JS: Cholesterol, cancer, and public health policy. Am J Med 1983; 75: 494.

Stemmermann GN, Nomura AM, Heilbrun LK: Dietary fat and the risk of colorectal cancer. Cancer Res 1984; 44: 4633.

Takeya Y, Popper JS, Shimizu Y, Kato H, Rhoads GG, Kagan A: Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and Cali- fornia: incidence of stroke in Japan and Hawaii. Stroke 1984; 15:15.

Yaari S, Goldbourt U, Evan-Zohar S, Neufeld HN: Associa- tions of serum high-density lipoprotein and total cholesterol with total, cardiovascular and cancer mortality in a 7-year prospective study of 10,000 men. Lancet 1981; I: 1011.

Keys A: Alpha lipoprotein (HDL) cholesterol in the serum and the risk of coronary heart disease and cancer. Lancet 1980; II: 603.

878 May 1986 The American Journal of Medicine Volume 60

Date post:	19-Oct-2016
Category:	Documents
Upload:	dwayne-reed
View:	214 times
Download:	1 times

Lipids and lipoproteins as predictors of coronary heart disease, stroke, and cancer in the honolulu...

Documents