Association of Risk Factor Variables

Transcription

1 Association of Risk Factor Variables and Coronary Artery Disease Documented with Angiography DAVID R. HOLMES, JR., M.D., LILA R. ELVEBACK, PH.D., ROBERT L. FRYE, M.D., BRUCE A. KOTTKE, M.D., PH.D., AND RALPH D. ELLEFSON, PH.D. SUMMARY Stepwise linear discrimination was used to analyze risk factors in 431 consecutive patients who underwent coronary angiography to determine which variables were most closely associated with coronary artery disease. Twenty-one risk factors were considered: total plasma cholesterol and triglycerides; the cholesterol and triglyceride content of high-density lipoproteins (HDL), low-density lipoproteins (LDL) and very low density lipoproteins (VLDL); and the percentage of total cholesterol and triglycerides in each fraction. Age, smoking history, family history, hypertension, diabetes mellitus and relative weight were also considered. Coronary artery disease was assessed using three standard grading scores. There were significant differences in risk factors between males and females. In males, LDL cholesterol and age were selected by multivariate analysis. In females, the ratio of HDL cholesterol to total cholesterol, as well as relative weight, family history, age and smoking were selected. The discriminating value of HDL cholesterol as the percentage of total cholesterol was significantly greater than that of HDL cholesterol itself. Despite highly significant associations between risk factors and the presence of coronary artery disease, the discrimination did not provide sufficient separation of the groups to give results that are useful diagnostically in individual patients. THE RELATIONSHIP between hyperlipidemia and an increased risk of coronary heart disease has been well documented and has served as a motivating factor for research into lipoprotein structure, function and metabolism. The aim of the research has been to isolate lipoprotein factors that are most closely associated with coronary heart disease in order to gain some knowledge about the pathophysiology of atherosclerosis and to allow clinicians to identify persons who are at high risk of developing the disease. Although Barr et al.' reported in 1951 the inverse relationship between high-density lipoproteins (HDL) and the incidence of coronary heart disease, until recently most attention has been directed at two of the principal carriers of plasma cholesterol, very low density (VLDL) and low-density (LDL) lipoproteins. During the past several years, however, the importance of HDL has been widely recognized,2 9 and in some studies, HDL appears to have been most predictive of coronary artery disease.10 Those findings support experimental evidence that HDL may have a role in inhibiting the accumulation of cholesterol in arteries by promoting the removal of cellular cholesterol."1-3 Other risk factors, including family history, hypertension, tobacco abuse, age, sex and diabetes also have been found to be associated with an increased risk of coronary artery disease.'4-16 Limited information is available concerning the predictive value From the Division of Cardiovascular Diseases and Internal Medicine, the Section of Medical Research Statistics and the Section of Clinical Chemistry, Mayo Clinic and Mayo Foundation, Rochester, Minnesota. Supported in part by research grant HL-14196, USPHS, NIH. Address for correspondence: D.R. Holmes, Jr., M.D., Mayo Clinic, 200 First Street SW, Rochester, Minnesota Received February 21, 1980; revision accepted June 17, Circulation 63, No. 2, of these variables and their relationship to angiographically defined coronary artery disease.6 ', The purpose of this study was to determine the relationship between angiographically defined coronary artery disease and some of the known risk factors, with specific attention to lipid fractions, and to determine the value of risk factors in predicting the presence of coronary artery disease in patients who undergo coronary angiography. Materials and Methods Four hundred thirty-one consecutive patients (336 males and 95 females) who underwent cardiac catheterization for evaluation of suspected coronary artery disease from 1975 to 1976 were evaluated. Twenty-one independent variables were considered: Total cholesterol;* the cholesterol in VLDL, LDL and HDL; and the ratios of the three fractions to the total; Total triglyceride;* the triglyceride in VLDL, LDL and HDL; and the ratio of the three fractions to the total; Age, relative weight, hypertension (yes or no), smoking (current, past or never), pack-years, diabetes (yes or no), and family history of coronary heart disease.t Coronary angiography was performed by the Judkins or Sones technique. Multiple views of the right and left coronary arteries were recorded. The angiograms were reviewed and coded by two car- *Whole-serum cholesterol and triglyceride were expressed as relative deviates from sex-specific regression of the natural logarithm on age as determined in 458 healthy persons. ta positive history of angina pectoris, myocardial infarction, or sudden abrupt death without obvious qause in the patient's direct blood relatives before age of 55 years. Blood relatives include the patient's parents, siblings, aunts, or uncles related by blood.

2 294 CI RCULATION VOL 63, No 2, FEBRUARY 1981 diologists without knowledge of the patient's lipid levels. The presence and severity of coronary artery disease was determined by consensus. The coronary circulation was divided into 27 segments and the degree of stenosis was coded for each segment. Blood samples were drawn before coronary angiography. Total plasma cholesterol was determined by a semiautomated colorimetric method,20 and triglyceride levels were determined by a semiautomated fluorometric method.20 Lipoprotein electrophoresis and isolation and quantitation of the LDL, VLDL and HDL were performed using techniques previously described The precision of these methods has been verified on blind-duplicate samples. To identify risk factors associated with the presence, severity and extent of coronary artery disease discovered at angiography, three methods were used: (1) stepwise linear discrimination between the two subgroups: those with significant coronary artery disease and those without significant coronary artery disease (i.e., the presence of any segment with > 70% stenosis of the luminal diameter), and (2 and 3) stepwise regression to predict the severity and extent of coronary artery disease, as measured by the Jenkins et al.9 scores and by the Gensini22 scores. Results Characteristics of the Study Group Of the 431 patients, 336 were males and 95 were females. The distributions of the lipids and lipid fractions in the study group are given by sex in table 1. The females had a smaller percentage of current and past smokers (48% vs 81 %) and a higher prevalence of hypertension (52% vs 34%). There were no significant sex differences in the prevalence of diabetes, prior angina or prior myocardial infarction (table 2). Eighty-eight percent of the males had significant coronary artery disease (> 70% luminal narrowing), compared with 63% of the females. The group without significant coronary artery disease was younger and had a lower prevalence of prior angina* and myocardial infarction.t Relationship of Lipids and Results of Arteriography For each sex, the quartiles of the distribution of each of the lipid fractions were determined. For each of these quartiles, the arteriographic results were summarized in terms of the severity of coronary artery disease, as judged by a mean Jenkins score (table 3). In males, the Jenkins score increased by 35% (7.86 to 10.59) from the lowest to the highest quartile of LDL cholesterol. In females, the mean Jenkins score decreased by 59% (8.43 to 3.46) from the lowest to the highest quartile of the ratio of HDL cholesterol to total cholesterol. *The diagnosis of angina pectoris was based on a history of exertional substernal chest discomfort. ta myocardial infarction was considered to have occurred if the patient stated that he had been hospitalized or had been told by his personal physician that he had suffered a myocardial infarction. Stepwise Linear Discrimination (table 4) In males, LDL cholesterol and age were selected as the variables that had significant discriminating power between those with and those without significant coronary artery disease. In addition, total cholesterol and the ratio of LDL cholesterol to total cholesterol were significant on univariate analysis. In males younger than 55 years of age, total cholesterol and pack-years were selected. In this group, as in all males, total cholesterol and LDL cholesterol were essentially tied for first place. In the older group, no variables were significantly associated with the presence of disease. In females, the ratio of HDL cholesterol to total cholesterol was the first variable selected; also selected were age, smoking, relative weight and family history. Seven other variables were significant by univariate methods, and of these, the first was the ratio of HDL triglyceride to total triglyceride, which was positively correlated with the ratio of HDL cholesterol to total cholesterol (r = 0.51). The ratio of HDL cholesterol to total cholesterol was of major importance in both female age groups. In those younger than 55 years, this ratio was essentially equal with family history, which was selected as the first variable. Smoking was the third variable. Pack-years and the ratio of HDL triglyceride to total triglyceride also were significant (p < 0.01). In women age 55 years or older, the ratio of HDL cholesterol to total cholesterol and relative weight were selected. No other variables were significant. The results of the stepwise regression for the Jenkins et al.9 and Gensini22 severity and extent scores reproduced those given above, except for minor differences in the ordering of variables. In females, the discriminating value of HDL cholesterol expressed as a percentage of total cholesterol was much greater than that of the HDL cholesterol concentration itself. The significance level for the ratio was < 0.001, compared with the concentration itself, which was of only borderline significance (p < 0.05). Even with the use of the ratio of HDL cholesterol to total cholesterol, there was considerable overlap between females with and without disease (fig. 1). In males, HDL did not discriminate between those with and those without disease. Even with the use of LDL, the overlap between males with and without disease was considerable (fig. 1). While the discriminant functions served to separate the mean scores for subjects with and without significant coronary disease, the overlap in the distributions was substantial, and only 65% of the males and 76% of the females could be classified correctly (fig. 2). Discussion Since an association of cholesterol and the development of atherosclerosis was recognized in 1913,23 an association of lipoprotein abnormalities and ischemic heart disease has been well documented. Although LDL cholesterol, total cholesterol and total triglyceride levels are directly related to prevalence of coronary artery disease, HDL cholesterol has been

3 RISK FACTORS AND CAD/Holmes et al. 295 TABLE 1. Distribution of Lipids and Lipid Fractions (mg/dl) in the Study Group Percentiles of the distribution Range P10 P25 P50 P75 P90 Cholesterol Total M* F* HDL M F LDL M F VLDL M F Triglyceride Total M F HDL M F LDL M F VLDL M F *M = 336; F 95. Abbreviations: HDL density lipoprotein. high-density lipoprotein; LDL low-density lipoprotein; VLDL - very low TABLE 2. Characteristics of the Study Group of Patients Undergoing Coronary Angiography Males Females <707O * >70%* <70%* >70%* (n= 40) (n= 296) (n= 35) (n= 60) Characteristic No. % No. % No. % No. % Age (years) < HistorY of angina History of myocardial infarction History of diabetes History of hypertension Family history Smoking CUrrent Past Never *ReferS to maximal stenosis.

4 296 CIRCULATION VOL 63, No 2, FEBRUARY 1981 TABLE 3. LDL chol HDL chol HDL chol/total chol VLDL chol Trig total LDL trig HDL trig Jenkins Angiography Score for Quartiles of the Lipid Distributions by Sex Lipid Q_ Q2 Q4 Q3 Mean SEM Mean SEM Mean SEM Mean SEM Chol total M F M F m F M F M F M F M F M F a VLDL trig M F Abbreviations: Chol - cholesterol, LDL = low-density VLDL very low density lipoprotein; Trig - triglyceride. X.tJ9 U a a lipoprotein; HDL = high-density lipoprotein; TABLE 4. Results of Stepwise Linear Discrimination and Univariate Analysis of Risk Factors in Patients Undergoing Coronary Angiography Males Females <55 >55 <55 >55 Total years years Total years* years* Patients Patients with > 70,rC. st.enosis of coronary arteries Variables selected by LDL Total None R-C-HI)L Family R-C-HDL stepwise linear cholesterol cholesterol history discrimnination Age Pack-years Relative weight Smoking Relative Age Family history Smoking weight Additional variables Total LDL None R-T-HDL R-C-HDL None significant (p < 0.05) cholesterol cholesterol on univariate analysis R-C-LDL R-T-VLDL Relative Total triglyceride weight LDL triglyceride Pack-years HDL cholesterol Total LDL cholesterol cholesterol VLDL cholesterol *Because of the few females, only the five variables selected for the total female group were used in the two age groups. Abbreviations: LI)L = low-density lipoprotein; HDL = high-density lipoprotein; VLDL = very low density lipoprotein; R-C-HDL or R-C-LDL = ratio of fraction to total cholesterol; R-T-HDL or R-T-VLDL ratio of fraction to total triglyceride.

5 RISK FACTORS AND CAD/Holmes et al. 297 Q J (D C.. 0) Females A /\ > 70% 1 %X "-- < 70% 1, 1 30[ 201 io0 336 Males - '70% n= <70% n= ua) CL ti N l HDL-chol/total chol x 100 Males - >70% ---<70% LDL chol (mg/dl) FIGURE 1. Distribution of ratio of high-density lipoprotein cholesterol (HDL chol) to total cholesterol (percent) for 60 females with maximal stenosis > 70% and 35 with maximal stenosis < 70%, and distribution of low-density lipoprotein cholesterol (LDL chol) in 295 males with maximal stenosis 70% and 39 with maximal stenosis < 70%. found to be more closely associated.2 An inverse association between HDL cholesterol and coronary artery disease persists even when adjusted for LDL cholesterol and triglyceride levels.2 Some investigators have found that the HDL cholesterol level correlates with the extent and severity of coronary artery disease9 and may be correlated with specific anatomic patterns.5 Although the total HDL cholesterol is better in predicting coronary artery disease than total cholesterol, some authors have found that the ratio of HDL cholesterol to total cholesterol may be even more useful.3 6 In our population, there were significant differences between males and females in regard to lipid fractions most closely associated with coronary artery disease. In females, the best correlation was the ratio of HDL cholesterol to total cholesterol, which was significantly better than the HDL cholesterol level itself. This was true in both age groups when either a continuous variable (that is, extent and severity of coronary artery disease) or the dichotomous variable (the presence or absence of coronary artery disease) was used as an end point. Seven additional lipid fractions were significant in univariate analysis, including HDL cholesterol and LDL cholesterol (table 4). In 0 CL i_ 95 Females ' 70% n % n-35 \, X/ 1 I Discriminant score FIGURE 2. Discriminant function scores for males and females with and without coronary artery disease. For males, the score is based on age and low-density lipoprotein cholesterol, for females, the score is based on R-C-HDL (ratio of high-density lipoprotein fraction to total cholesterol), relative weight, age, family history of coronary artery disease and smoking. males, the only lipid variable selected was LDL cholesterol, although total cholesterol and the ratio of LDL cholesterol to total cholesterol were significant on univariate analysis. This was also most significant regardless of whether the end point was the presence or absence of coronary artery disease or the extent and severity of coronary artery disease. In contradistinction to the results in women, we could not identify any significant risk factors in men age 55 years or older. This may indicate that aging and sex are such powerful risk factors that they "wash out" the effect of the other variables.24 In contrast to the findings of some other investigators,5' 9 we found no significant relationship between HDL cholesterol and coronary artery disease in males. In contrast to Pearson et al.,5 who found that low levels of HDL cholesterol were associated with left main coronary artery disease, we found that the risk factors associated with left main coronary artery disease were similar to the variables associated with coronary artery disease as a whole, that is, LDL cholesterol and age. The reasons for these differences are unclear but may relate to different patient populations. The mode of action of HDL is not fully understood. Biochemical studies have shown that the body

6 298 CI RCULATION VOL 63, No 2, FEBRUARY 1981 cholesterol pool is negatively correlated with the concentration of plasma HDL cholesterol.'1 Miller and Millerll and Stein et al.13 suggested that HDL cholesterol facilitates the removal of intracellular cholesterol from nonhepatic tissue and functions as a transport vehicle, returning the cholesterol to the liver for catabolism and excretion. HDL also might inhibit the entrance of LDL into cells by competitive binding with the LDL receptors.'3 Either mechanism may have an important role in the development of atherosclerosis. While the entry of cholesterol into the cells is facilitated by the binding of LDL to surface receptors,25 the rate of LDL uptake is regulated by the cellular content of cholesterol. Homeostasis of cellular cholesterol seems to be maintained by a balance between uptake and removal mechanisms, and in turn, the cellular level of cholesterol appears to influence cellular cholesterol synthesis. The relationships among LDL binding, cholesterol uptake, cholesterol transfer to HDL from cells, LDL levels and dynamics, HDL levels and dynamics, and cholesterol biosynthesis all may be significant in the development of atherosclerosis.25' 26 In addition to differences between males and females in lipid fractions most closely associated with coronary artery disease, there were also sex-related differences in the other risk factors. In both young males and females, smoking history also was selected on linear regression. This is similar to other reported series, which have documented that the younger the age group, the higher the risk associated with smoking.24', 2 28 The other variables selected with linear regression in the females were family history, relative weight and age. In the younger age group, the family history of early onset of arteriosclerotic heart disease was the strongest variable. This supports evidence that has demonstrated an increased risk of coronary artery disease in relatives of people who have the onset of coronary artery disease at an early age Patients with positive family histories were younger and had higher cholesterol levels and lower ratios of HDL cholesterol to total cholesterol. Hypertension has been found to be an important risk factor for the development and prognosis of coronary artery disease. The fact that hypertension does not seem to discriminate at the angiographic level in this and other studies18 may be due to the weakness of the measure used, that is, a history of the presence or absence of hypertension. If we could measure the total impact in terms of the duration and degree of elevation of the blood pressure, the presence of hypertension might well be found to be more significant. Studies on the association of risk factors and coronary artery disease have used several different markers of the disease. Some of those have had significant limitations. Those studies based on the clinical diagnosis of coronary artery disease have been by their very design limited because certain patient populations, for example women, may have clinical coronary artery disease with angina pectoris without having angiographically demonstrable coronary artery disease. More recent studies that have used coronary angiography to define the population samples have solved this problem but still have had limitations. The use of dichotomous variables such as stenosis of less than 70% and of 70% or more in identifying those factors associated with the presence and severity of coronary artery disease has some potential problems. These can be avoided by using a continuous variable end point; for example, scores that consider both the severity and extent of disease. In our study, however, the significant risk factors were similar, regardless of which coronary score was used. Risk factor studies in a population that undergoes coronary arteriography have the advantage that the major proportion of the misclassification present in clinically diagnosed series is clearly defined and provides a comparison group. This advantage, however, is offset by the complex and largely undefined process by which the study group is selected. The angiographic group, as well as the subset in which no coronary artery disease is found is not representative of the population at large. The selection process that results in a population that undergoes coronary angiography may significantly alter (intensify or dilute) the associations in the population as a whole between coronary artery disease and various risk factors. Even in our angiographic series, although highly significant relationships were found, they were not clinically useful at the level of an individual patient in predicting coronary artery disease. References 1. Barr DP, Russ EM, Eder HA: Protein-lipid relationships in human plasma. II. In atherosclerosis and related conditions. Am J Med 11: 480, Castelli WP, Doyle JT, Gordon T, Hames CG, Hjortland MC, Hulley SB, Kagan A, Zukel WJ: HDL cholesterol and other lipids in coronary heart disease: The Cooperative Lipoprotein Phenotyping Study. Circulation 55: 767, 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 62: 707, Gofman JW, Young W, Tandy R: Ischemic heart disease, atherosclerosis, and longevity. Circulation 34: 679, Pearson TA, Bulkley BH, Achuff SC, Kwiterovich PO, Gordis L: The association of low levels of HDL cholesterol and arteriographically defined coronary artery disease. Am J Epidemiol 109: 285, Williams P, Robinson D, Bailey A: High-density lipoprotein and coronary risk factors in normal men. Lancet 1: 72, Miller NE, Forde OH, Thelle DS, Mjos OD: The Troms0 heart-study: high-density lipoprotein and coronary heartdisease; a prospective case-control study. Lancet 1: 965, Ononogbu IC: High density lipoproteins in ischaemic heart disease. Experientia 33: 1063, Jenkins PJ, Harper RW, Nestel PJ: Severity of coronary atherosclerosis related to lipoprotein concentration. Br Med J 2: 388, Gordon T, Castelli WP, Hjortland MC, Kannel WB, Dawber TR: Diabetes, blood lipids, and the role of obesity in coronary heart disease risk for women: The Framingham Study. Ann Intern Med 87: 393, Miller GJ, Miller NE: Plasma-high-density-lipoprotein concentration and development of ischaemic heart-disease. Lancet 1: 16, Carew TE, Koschinsky T, Hayes SB, Steinberg D: A mecha-

7 RISK FACTORS AND CAD/Holmes et al. 299 nism by which high-density lipoproteins may slow the atherogenic process. Lancet 1: 1315, Stein Y, Glangeaud MC, Fainaru M, Stein 0: The removal of cholesterol from aortic smooth muscle cells in culture and Landschutz ascites cells by fractions of human high-density apolipoprotein. Biochim Biophys Acta 380: 106, Reid DD, Hamilton PJS, McCartney P, Rose G, Jarrett RJ, Keen H: Smoking and other risk factors for coronary heartdisease in British civil servants. Lancet 2: 979, Rosenman RH, Brand RJ, Sholtz RI, Friedman M: Multivariate prediction of coronary heart disease during 8.5 year follow-up in the Western Collaboration Group Study. Am J Cardiol 37: 903, Gordon T, Sorlie P, Kannel WB: Coronary heart disease, atherothrombotic brain infarction, intermittent claudicationa multivariate analysis of some factors related to their incidence: Framingham Study, 16-year followup. In The Framingham Study: An Epidemiological Investigation of Cardiovascular Disease, edited by Kannel WB, Gordon T. Washington DC, US Govt Printing Off, 1971, Section Frick MH, DahWn G, Berg K, Valle M, Hekali P: Serum lipids in angiographically assessed coronary atherosclerosis. Chest 73: 62, Anderson AJ, Barboriak JJ, Rimm AA: Risk factors and angiographically determined coronary occlusion. Am J Epidemiol 107: 8, Welch CC, Proudfit WL, Sheldon WC: Coronary arteriographic findings in 1,000 women under age 50. Am J Cardiol 35: 211, Ellefson RD, Caraway WT: Lipids and lipoproteins. In Fundamentals of Clinical Chemistry, edited by Tietz NW. Philadelphia, WB Saunders Company, 1976, pp Ellefson RD, Elveback LR, Hodgson PA, Weidman WH: Cholesterol and triglycerides in serum lipoproteins of young persons in Rochester, Minnesota. Mayo Clin Proc 53: 307, Gensirti GG. Coronary Arteriography. Mount Kisco, New York, Futura Publishing Company, 1975, pp Anitschkow N, Chalatow S: Ueber experimentelle Cholesterinsteatose und ihre Bedeutung for die Entstehung einiger pathologischer Prozesse. Zentralbl AlIg Pathol 24: 1, Vlietstra RE, Frye RL, Kronmal RA, Sim DA, Tristani FE, Killip T III: Risk factors and angiographic coronary artery disease: a report from the Coronary Artery Surgery Study (CASS). Circulation 62: 254, Goldstein JL, Brown MS: The low-density lipoprotein pathway and its relation to atherosclerosis. Annu Rev Biochem 46: 897, Small DM: Cellular mechanisms for lipid deposition in atherosclerosis. N EngI J Med 297: 873, Hammond EC: Smoking in relation to the death rates of one million men and women. Nat Cancer Inst Monogr 19: 127, Hammond EC, Garfinkel L: Coronary heart disease, stroke, and aortic aneurysm: factors in the etiology. Arch Environ Health 19: 167, Epstein FH: Risk factors in coronary heart disease: environmental and hereditary influences. Isr J Med Sci 3: 594, Bloor CM: Hereditary aspects of myocardial infarction. Circulation 40 (suppl IV): IV-130, Bloor CM, McKusick VA: The genetics of coronary artery disease. In Symposium on Coronary Heart Disease, 2nd ed, edited by Blumgart HL. Dallas, American Heart Association, 1968, monograph no. 2, pp 6-11