Effect of Serum Lipids, Lipoproteins, and Apolipoproteins on Vascular and Nonvascular Mortality in the Elderly

The relationship between elevated serum total cholesterol and LDL-C concentrations, low HDL-C concentrations, and coronary heart disease in middle age¹²³⁴ and early old age has been established.⁵ In old age, however, the association between total cholesterol and coronary heart disease tends to disappear or even becomes inverse; therefore, the prognostic significance of cholesterol and lipoprotein fractions has remained controversial in this age group.^167891011 Several studies have suggested an association between low serum cholesterol and such diseases as cancer and chronic respiratory diseases^12131415 as well as certain lifestyle risk factors, such as heavy alcohol use and smoking.¹⁶¹⁷¹⁸ It has been proposed that the inverse association between cholesterol level and mortality from noncardiac causes is a result of these confounding factors.¹⁵¹⁶¹⁸ Because of this, it has been suggested that lipoprotein fractions and apolipoproteins can predict cardiovascular risk in the elderly more reliably than total cholesterol concentration.¹⁹²⁰²¹

Variation in apo E phenotype has been found to affect lipoprotein levels. Three common apo E alleles, e2, e3, and e4, have been identified. The e4 allele increases intestinal cholesterol absorption, affects LDL synthesis in the liver,²² and is associated with elevated levels of total cholesterol and LDL-C^23242526 and a higher prevalence of atherosclerosis.²⁷ The association between apo E polymorphism and coronary heart disease has been found in several cross-sectional populations.^28293031 Recently, results from the first longitudinal study (with 5-year follow-up) indicated that the e4 allele predicted death from coronary heart disease in elderly men.³² Nevertheless, information about the prognostic importance of the e4 allele in unselected populations is scarce.

The aim of this longitudinal study was to examine the prognostic impact of serum total cholesterol, LDL-C, HDL-C, triglyceride, and apo A-1, B, and E on death due to vascular and nonvascular causes in an unselected elderly population, taking into account the contribution of other risk factors at screening.

Methods

Subjects

In 1982 a large survey of the health status of the elderly was carried out in the city of Turku, Finland. A random sample of 480 community-living individuals aged 65 years and older, stratified into four age groups (65 through 69, 70 through 74, 75 through 79, and 80 years and older), with men and women grouped separately, was obtained from the register of the Social Insurance Institution. No exclusion criteria other than living in an institution were used. The participation rate was 72%; thus, the population consisted of 347 subjects (Table 1). Of these, 184 were men, and 163 were women. The reasons for nonparticipation were as follows: illness in 48 subjects, attitude in 50, social problems in 3, and unknown in 32. A full history was obtained and a complete clinical evaluation was performed in all 347 participants.

Clinical Examination

The clinical history was obtained by personal interview. A comprehensive clinical evaluation, including physical examination, standard electrocardiography, chest x-ray, blood pressure measurement, routine biochemical analysis, and weight and height measurements (expressed as body mass index), was carried out. Major vascular diagnoses were established on the basis of the history and clinical evaluation. Ischemic heart disease was diagnosed when there was a standard history of angina pectoris.³³ Myocardial infarction was recorded as present on the basis of Q wave abnormalities on the electrocardiogram, which were coded according to the Minnesota code.³⁴ Congestive heart failure was considered to be present when typical findings were found on chest x-ray. Cerebral artery disease was diagnosed if there was evidence of focal neurological symptoms or signs. Moreover, additional information was obtained from the national health insurance documents of the subjects. These documents include information on entitlement to preferential refund of medication expenses because of common chronic diseases. To get this entitlement, a complete clinical evaluation is needed. If a subject was eligible for refund of medication expenses because of diabetes, hypertension, ischemic heart disease, congestive heart failure, or chronic bronchitis with pulmonary emphysema, that diagnosis was recorded as present.

Chemical Analyses

Serum total cholesterol, HDL-C, LDL-C, triglyceride, and apo A-1, B, and E were measured using overnight fasting samples drawn in 1982. All other lipid parameters except apo E phenotyping were analyzed in 1982 on a daily basis after the samples were taken or after 2 to 3 months of storage at −70°C. Serum total cholesterol was determined by the enzymatic cholesterol esterase–cholesterol oxidase method (Boehringer Mannheim, Mannheim, Germany). HDL-C was measured after dextran–sulphate precipitation of serum³⁵ with the same enzymatic method. Serum triglycerides were determined with the enzymatic ultraviolet method (Boehringer Mannheim). The concentration of serum apo A-1 (reference values, 0.86 to 1.46 g/L) was determined by radioimmunoassay as described earlier.³⁶ Apo B (reference values 0.92 to 1.33 g/L)³⁶ was analyzed by immunoturbidometry according to Riepponen et al³⁷ (Orion Diagnostica, Espoo, Finland). Serum LDL-C was calculated using the Friedewald formula.³⁸ Apo E phenotyping was carried out from serum samples stored frozen (−40 C) by isoelectric focusing and immunoblotting after removal of serum lipids.³⁹

Follow-up

In 1994 the 11-year mortality rate of subjects and causes of deaths were obtained from the mortality statistics. One hundred ninety-nine (57%) of the participants had died, and 148 were still alive (Table 1). In 127 subjects, death was caused by vascular disease. In 54 (43%), diagnosis of vascular death was confirmed at autopsy. The vascular deaths were predominantly caused by coronary events, cerebral infarcts, or sudden cardiac death. In 72 subjects, death was due to nonvascular causes, and autopsy was carried out in 31 of these subjects (43%). The causes of death are given in Table 2. To determine the effect of blood lipids on the 11-year mortality rate, lipid values of subjects who were alive on January 1, 1994, and those of subjects who had died as a result of either vascular or nonvascular causes by that same date were compared.

Statistical Analysis

The SAS (Statistical Analysis System, version 6) program package was used for the analyses. The Student’s t test was used to compare mean values of normally distributed numerical variables. Product-limit (Kaplan-Meier) survival curves were calculated for various total cholesterol levels. The curves of these cholesterol data were compared using the log rank test. RRs (hazard ratios with 95% CIs) for the 11-year vascular and nonvascular mortality rates were calculated using the Cox proportional hazards model according to blood lipids, adjusting for other risk factors. Adjustment was made for age, sex, smoking, alcohol use, body mass index, coronary heart disease, hypertension, and diabetes. Age and body mass index were fitted continuously in the model, and sex, smoking (never, former, and current smoker), alcohol use (never, 1 to 2 drinks per week, and more than 2 drinks per week), and the presence of coronary heart disease (none, probable, or definite), diabetes (none or definite), and hypertension (none or definite) were fitted as categorical variables. Cholesterol was fitted at four levels (≤5.0, 5.1 to 6.5, 6.6 to 8.0, and >8.0 mmol/L); HDL-C, at three levels (<1.2, 1.2 to 1.7, and >1.7 mmol/L); LDL-C, at three levels (<3.5, 3.5 to 5.0, and >5.0 mmol/L); and triglyceride and apo E, at two levels (<2.0 and >2.0 mmol/L and e4 allele absent or present, respectively). Apo A-1 and B (g/L) were divided into tertiles and fitted at three levels (lowest, middle, and highest tertiles). To assess the adequacy of the Cox model, the interactions of age and other risk factors were tested with logistic regression. The Cox model proved to be adequate in all but the analysis of effect of HDL-C and apo A1 on nonvascular mortality, in which slight interactions between HDL-C and age (P=.05) and between apo A1 and age (P=.04) were observed. To examine the possible effects of preexisting diseases on mortality, deaths that occurred during the first 4 years of follow-up (1982 through 1985) were excluded from the further analysis of total cholesterol, and adjusted RRs were calculated using data from 1986 through 1993.

Results

The baseline characteristics of subjects grouped by mortality are given in Table 3. The mean concentrations of serum lipids, lipoproteins, and apolipoproteins in different age groups according to outcome are presented in Fig 1. Kaplan-Meier survival curves for vascular and nonvascular mortality, stratified by total serum cholesterol level, are shown in Figs 2 and 3. Low total cholesterol level was associated with death due to both vascular (P=.021) and nonvascular causes (P=.0027). Adjusted RRs according to lipids, lipoproteins, and apo A-1, B, and E are given in Table 4. After adjustment forother risk factors, the inverse association between total cholesterol and vascular mortality disappeared, but low total cholesterol was still a significant predictor of nonvascular death. RRs of death due to nonvascular causes for those with elevated total cholesterol (5.1 to 6.5, 6.6 to 8.0, and >8.0 mmol/L) compared with the reference group (≤5.0 mmol/L) were 0.5 (95% CI, 0.2 to 1.2), 0.6 (0.2 to 1.0), and 0.2 (0 to 0.8), respectively. Concentrations of HDL-C, LDL-C, triglyceride, or apo B were not associated with vascular or nonvascular mortality. On the other hand, a low concentration of apo A-1 predicted vascular death. The RR for the lowest tertile was 1.6 (1.1 to 2.5) compared with the highest tertile. When the mortality data from the first 4 years (1982 to 1985) was excluded from the further analysis of total cholesterol, the distribution of deaths did not change, indicating that the mortality rate obtained for the first 4 years did not differ from that obtained later.

The occurrence of the apo E e4 allele (Table 4) was associated with vascular mortality (RR, 1.5; 95% CI, 1.0 to 2.2), and a tendency of the e4 allele to predict nonvascular mortality was observed (RR, 1.5; 95% CI, 0.9 to 2.5). The adjusted RRs of vascular and nonvascular death according to the presence of the apo E e4 allele were analyzed separately for each age group (Table 5). A significant increase in risk was observed in two age groups, 70 to 74 and 75 to 79 years. The distribution of apo E phenotypes and gene frequencies grouped by mortality are shown in Table 6. The unadjusted figures of phenotypes or gene frequencies of alleles e2, e3, and e4 did not differ between subjects who were alive or those who had died. The mean concentrations of lipids, lipoproteins, and apo A-1 and B in subjects with and without the apo E e4 allele are shown in Table 7. The concentrations of total cholesterol, LDL-C, and apo B were significantly higher in the group with the e4 allele. The concentrations of HDL-C, apo A-1, or triglyceride did not differ according to the presence of an apo E allele.

Discussion

An elevated total serum cholesterol level did not predict vascular mortality but was a strong predictor of survival from death due to nonvascular causes. There are several possible explanations for the absence of a positive association between total cholesterol and vascular mortality. First, the total cholesterol level in the elderly may not represent their lifetime exposure, because many lifestyle factors and diseases may modulate it with advancing age.¹¹ Thus, total cholesterol values measured in old age may differ considerably from levels determined by genetic and dietary factors in young and middle age. Second, it is possible that the elderly study population was already selected at baseline according to cholesterol level and that those who remained in the cohort may have been relatively resistant to the effects of lipids. It is known that the reference values of serum total cholesterol do not increase with advancing age in men but do in women.⁴⁰ Third, a selection bias may have been caused by the nonparticipation of individuals with a low total cholesterol level and good health. In the present study, the inverse association between total cholesterol and nonvascular mortality was probably caused by the confounding of secondary reduction of total cholesterol concentration due to age-related lifestyle factors or preexisting diseases. This inverse association was further analyzed by excluding the first 4 years of mortality data after the screening and analyzing the years 1986 through 1993 separately. Nevertheless, the inverse association was still detected, suggesting that chronic diseases with a relatively long-term prognosis rather than disseminated cancer or other diseases with a poor short-term prognosis were responsible for the deaths. The negative association between total cholesterol and cancer mortality may persist for 18 years after screening, even when deaths occurring during the early follow-up period are excluded.⁴¹⁴² Our findings confirm that total cholesterol concentration is sensitive to many factors, making it an unreliable marker for estimating cardiovascular risk in the elderly. However, the conclusion that a high total cholesterol concentration protects against vascular or nonvascular mortality cannot be drawn.¹⁵

It has been suggested that LDL-C indicates cardiovascular risk in elderly individuals better than total cholesterol.⁴³ A recent study provides evidence that low HDL-C is associated with increased coronary heart disease mortality in older age groups.²¹ Furthermore, low values of LDL-C and high values of HDL-C have been found in octo-and nonagenarian survivors.⁴⁴ In the present study, however, no association between LDL-C or HDL-C and vascular mortality was found.

Little data exist on the importance of apo A-1 as a vascular risk factor among the elderly. Longitudinal studies do not exist. Patients with myocardial infarction had lower values of apo A-1 compared with controls in the study of Avogaro et al.²⁰ On the other hand, apo A-1 was not associated with coronary heart disease in a cross-sectional study of elderly men.⁴⁵ In our popuation, the low concentration of apo A-1 was the strongest predictor of vascular death, whereas HDL-C showed no prognostic significance. In survivors as well as nonsurvivors, the concentration of apo A-1 was lower than previously reported in Finland in the middle-aged population²⁴ and among elderly men.⁴⁵

In cross-sectional samples, the apo E e4 allele has been found to increase the risk of coronary heart disease.³⁰³¹⁴⁶ Also, the high prevalences of the e4 allele²⁴ and coronary heart disease in Finns and the low prevalences of the e4 allele and coronary heart disease in Chinese⁴⁷ and Japanese⁴⁸ populations support the concept of the e4 allele as an important indicator of atherosclerotic disease. In the study of Stengård et al, the unadjusted analysis of elderly men showed that the e4 allele predicted excess mortality due to coronary heart disease. In our study, the apo E allele frequencies were comparable with frequencies in the Finnish population, including youths.²⁴³²³⁹ Adjusted vascular mortality was 50% higher in subjects with the e4 allele than in those without the e4 allele. The association was strongest in the 70- to 79-years age group but nonsignificant in subjects aged 80 or more years. However, there were only 8 survivors in the highest age group. In subjects having the e4 allele, total cholesterol, LDL-C, and apo B concentrations have been found to be higher than in subjects with other alleles.²⁴²⁶ Also, in the present study, total cholesterol, LDL-C, and apo B concentrations were significantly higher in subjects with the e4 allele than in those without the e4 allele. The lack of association between total cholesterol and LDL-C and vascular mortality, however, suggests that the risk was not related to dyslipidemia. It is possible that the cardiovascular risk associated with the e4 allele is mediated not only by dyslipidemia but also by other mechanisms. Apo E is involved in the immune system and tissue regeneration, which makes possible a direct contribution to the atherogenic process at the cellular level in the arterial wall.⁴⁹⁵⁰ Moreover, the association between the e4 allele and Alzheimer’s disease suggests that unknown mechanisms, unrelated to serum lipids but involving the risk of atherosclerosis, may exist.

When cardiovascular risk due to dyslipidemia is evaluated, the selection of subjects is essential because lipids, especially total cholesterol, also reflect other factors as inheritance and diet. In the present study, random selection was used, and other risk factors were controlled by means of multivariate analysis. The response rate was 72%, which is not as high as one might wish. It is, however, comparable with that of studies including ambulatory electrocardiographic monitoring and comprehensive evaluation of health status. In this kind of study, cooperation and a positive attitude are needed. The unselected elderly study population also included a number of frail individuals. The low response rate may have caused some selection bias since the frail and sick individuals could have been those who did not respond and whose blood lipids underwent secondary changes. In the present study, however, the number of subjects who did not attend because of illness was considerably small, around 10% of the sample. In addition, the inverse association between total cholesterol and mortality due to nonvascular causes suggests that this kind of selection was unlikely. Rather, individuals with low total cholesterol level and good health or individuals with elevated total cholesterol and poor short-term prognosis did not attend. The prevalences of preexisting diseases were comparable with those in the large Finnish population study with age groups of 65 to 99 years carried out in 1978 to 1981.⁵¹ This also suggests that the significant selection bias did not take place in our study. Although the sample size was smaller than in some previous cohort studies, a large number of deaths occurred during the long follow-up period, yielding sufficient statistical power. The methods for the determination of apo A-1 and B were new in 1982, and one can propose that the standardization of the methods was not accurate at that time. However, the levels of apolipoproteins presented in our study are comparable with those observed in other Finnish studies, suggesting that our results are valid.³⁶⁵²

In conclusion, in this cohort of elderly people, the predictive value of blood lipids differed from that of middle-aged people. The prognostic impact of elevated total cholesterol as a cardiovascular risk factor disappeared with aging but predicted survival in subjects with nonvascular diseases. HDL-C and LDL-C had no pre-dictive value. The presence of the apo E e4 allele and a low concentration of apo A-1 proved to be the only predictors of vascular mortality. The risk associated with the apo E e4 allele was not related to dyslipidemia. Thus, the apo E e4 allele and a low concentration of apo A-1 are obviously more reliable indicators of cardiovascular risk in elderly people than lipoprotein fractions or total cholesterol, because the latter are affected by age-related confounding factors.

Selected Abbreviations and Acronyms

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Footnotes