Factors Associated with Lipoprotein Cholesterol Levels

Transcription

1 Factors Associated with Lipoprotein Cholesterol Levels The Framingham Study Peter W. F. Wilson, Robert J. Garrison, Robert D. Abbott, and William P. Castelli Lipoprotein cholesterol determinations in 6328 individuals aged 20 to 79 years in the Framingham Heart Study Cohort and Offspring were related by multiple regression analysis to reported cigarette smoking, alcohol intake, and (wf ht 2 ). Cigarette smoking was found to be strongly associated with "atherogenic" lipoprotein cholesterol profiles in young adults, and particularly in women. The associations for alcohol intake were mostly uniform across age s and lipoprotein cholesterol fractions, while coefficients for varied considerably. Blood chemistry associations were studied in participants under 50 years, after controlling for smoking,, and alcohol intake. Significant associations for high density lipoprotein cholesterol (HDL-C) were seen for both sexes with alkaline phosphatase, serum calcium, serum uric acid, and leucocyte count. Low density lipoprotein cholesterol (LDL-C) associations were observed with calcium, hematocrit, lactate dehydrogenase, and leucocyte count in men and women, while very low density lipoprotein cholesterol (VLDL-C) associations occurred with alkaline phosphatase, glucose, uric acid, and leucocyte count in both sexes. A 1 mg/dl higher calcium corresponded to an HDL-C approximately 4 mg/dl greater and a LDL-C typically 6 mg/ dl greater after controlling for 12 other variables. A 1000/ml increase in leucocyte count was typically associated with a decrease in HDL-C by 1 mg/dl and an increase in LDL-C and VLDL-C of 1 mg/dl each. (Arteriosclerosis 3: , May/June 1983) Recent research strongly suggests that lipoprotein metabolism is intimately linked to the atherosclerotic process. These findings prompted numerous investigators to study the biological concomitants of lipoprotein measurements. 1 " 3 Most of these reports have concentrated on only a few variables at a time and dealt with relatively small populations. In the Framingham Heart Study, lipoprotein cholesterols were determined in individuals aged 20 to 79 years during 1971 to An extensive medical history was taken and physical examination performed, which allowed us to examine the multivariate association of obesity, alcohol, cigarette smoking, and drug intake with lipoprotein cholesterol levels for men and women whose ages spanned six decades. Automat- From the Epidemiology and Biometry Program, National Heart, Lung, and Blood Institute, Bethesda, Maryland, and the Heart Disease Epidemiology Study, National Heart, Lung, and Blood Institute, Framingham, Massachusetts. Address for reprints: Dr. Peter W. F. Wilson, National Heart, Lung, and Blood Institute, Federal Building, Room 300, Bethesda, Maryland Received July 27,1982; revision accepted December 7, ed chemistry and hematology tests were performed for the younger one-half of the population. This additional information provided the basis to investigate potential relationships between individual lipoprotein cholesterols and these other factors. Methods The population studied included 2217 men and women from the Framingham Heart Study Cohort which has been followed with biennial exams since 1949, and 4197 men and women from the Framingham Offspring Study at the time of their first examination from 1971 to The Cohort participants in this report include men and women aged 50 to 79 years old who were free from coronary heart disease at the time of the eleventh biennial examination. Women taking estrogen preparations were excluded from all analyses. The Offspring participants included men and women aged 20 to 49 years, subject to the same restrictions as the Cohort. The data collection for both the Cohort and Offspring s consisted of a full medical history,

2 274 ARTERIOSCLEROSIS VOL 3, No 3, MAY/JUNE 1983 physical examination, and fasting lipoprotein cholesterol determinations using a modified Lipid Research Clinics (LRC) protocol. 4 Current cigarette smoking history and the usual intake of cans of beer, glasses of wine, and cocktails per week were included in the medical history. The weekly alcohol intake was calculated for each individual using the following equation: intake (oz/wk) = (weekly number of 12 oz cans of beer x 0.444) + (weekly number of 4 oz glasses of wine x 0.400) + (weekly number of 1.5 oz liquor cocktails x 0.570). 5 The was computed for each individual by dividing the subject's weight in kilograms by his or her height in meters raised to the second power. A manual Abell-Kendall method was used to measure cholesterol, 6 and triglycerides were determined by the AAI modification of the Kessler-Lederer method. 7 High density lipoprotein cholesterol (HDL- C) was measured after precipitation of other lipoproteins with heparin-manganese. Low density lipoprotein cholesterol (LDL-C) was determined by subtracting HDL-C from the bottom fraction cholesterol, performed after reconstitution of the bottom fraction following preparative ultracentrifugation of EDTA plasma at serum density. The very low density lipoprotein cholesterol (VLDL-C) was calculated by subtracting the bottom fraction cholesterol from total cholesterol. 4 If the calculated VLDL-C value was negative, it was arbitrarily set at zero. Automated blood chemistries (SMA 12/60) and hematology tests using a Coulter Counter were performed on the Offspring at the Framingham Union Hospital clinical chemistry laboratory. Statistical methods included use of a standard multiple linear regression model using least squares estimation for Tables A multivariate regression model derived from a backward selection procedure generated Tables Factors used in this analysis included current cigarette smoking (yes or no),, alcohol intake per week, serum albumin, serum alkaline phosphatase, blood urea nitrogen (BUN), serum calcium, serum globulin, plasma glucose, lactate dehydrogenase (LDH), serum phosphorus, serum glutamic oxaloacetic transaminase (SGOT), serum uric acid, blood hemoglobin, and leucocyte count. Results The mean values and their standard deviations for the frequency of smoking,, and alcohol intake are displayed in Table 1 for men and Table 2 for women. Among the men, each variable shows Table 1. Mean Values and Standard Deviations for Frequency,, and Intake In Framingham Men Years 40-^9 Number Current smoking Proportion Mean (kg/sq m) Mean intake (oz/wk) Table 2. Mean Values and Standard Deviations for Frequency,, and Intake In Framingham Women Years qroup Number Current smoking Proportion Mean (kg/sq m) Mean intake (oz/wk)

3 ASSOCIATIONS WITH LIPOPROTEIN CHOLESTEROLS Wilson et al. 275 different cross-sectional trends with age. frequency decreased, rose at least through the sixth decade, and alcohol intake peaked in the 40 to 49 year age. Table 2 shows that the smoking frequency for women paralleled that for the men, while the increased gradually over the six decades examined. Reported alcohol intake for the women was less than one-half that of the men, with a maximum intake in the middle age s. The results of multiple regression analyses for each of the lipoprotein cholesterol fractions are shown in Tables 3 to 12. The gender-specific analyses are presented for each 10-year age and for the total sample in Tables 3 to 8. The estimated multiple regression coefficients for HDL-C appear in Table 3 for men and Table 4 for women. From an inspection of the coefficients for the total male population, current smoking was associated with an HDL- C that was 3.2 mg/dl lower in nonsmokers. A significant inverse association was also found for the in all s. Overall, each unit increment in this index of adiposity was accompanied by a 0.8 mg/dl decrement in HDL-C. exerted its effect in the opposite direction. Each additional ounce of alcohol consumed showed a concomitant 0.4 mg/dl increase in HDL-C. Analysis by 10-year age conforms to the pattern discussed for the total population. The strength of associations indicated by the multiple regression coefficient estimates is seen among most of the age s, with uniformity in the estimates the rule, despite the relatively small values of the multiple values, which range from 0.08 to The results in Table 4 paralleled those for men, with few exceptions. The individual variables manifested their effect in the same direction as before, but the magnitude of the coefficients differed. The coefficients were very similar in men and women, but appeared to be slightly larger for smoking, and approximately twice as large in women for alcohol. Women who smoke tended to have HDL-C levels 5.0 mg/dl less than their nonsmoking peers. For each ounce of alcohol consumed, the HDL-C for these women was 0.9 mg/dl greater. Table 5 and 6 present the regression results for LDL-C. The low values for the multiple s, especially for individuals older than 40 years, demonstrates that prediction of LDL-C with this model is quite weak. For the total male and female popula- Table 3. Regression Coefficients of Selected Variables for HDL Cholesterol by 10-Year Group In Framingham Men 66.64: 62.1* 58.7* 72.7* 64.3* * * < : p < to.001 < p <. *p< * -3.8* -2.4f -2.7* * -0.9* -0.7* -0.6* -1.1* -0.8* -1.6* -0.8* 0.6* 0.5* 0.5* 0.5* 0.7* 0.8* 0.4* 0.16* 0.11* 0.10* 0.16* 0.11* 0.08* 0.12* Table 4. Regression Coefficients of Selected Variables for HDL Cholesterol by 10-Year Group In Framingham Women 67.7* 71.9* 80.9* 77.8* 79.5* 81.1* 72.4* t0.001 < p<. -4.3* -4.6* -6.2* -5.9* * -0.5* -0.7* -0.9* -0.7* -0.9* -1.0* -0.6* 0.5f 0.9* 1.1* 1.0* 1-0t * 0.07* 0.10* 0.14* 0.10* 0.09* 0.09* 0.08*

4 276 ARTERIOSCLEROSIS VOL 3, No 3, MAY/JUNE 1983 Table 5. Regression Coefficients of Selected Variables for LDL Cholesterol by 10-Year Group In Framingham Men 67.3* 96.9* 135.8* 123.0* 115.9* 149.0f 99.8* * < p < t0.001 < p< * 1.3* * -0.7* * 0.02* Table 6. Regression Coefficients of Selected Variables for LDL Cholesterol by 10-Year Group In Framingham Women 88.4* 79.0* 95.2* 149.6* 140.4* 154.7* 81.8* * < p < fo.001 < p< * 7.9t * 1.4* 1.3* * -1.2f * t 0.03f 0.06* 0.05* 0.02* 0.06* tions, the was positively associated with levels of LDL-C, but these associations were found only in the youngest male and female age s. was positively associated with LDL-C in women younger than 60 years, and a suggestion of a negative relationship was present in the older s. A similar pattern was seen for the men. intake also showed a largely negative association with LDL-C for women. This association was significant overall for women, but not for men. Although no statistically significant association was found, the data suggest an inverse relationship between alcohol intake and LDL-C for men as well. Tables 7 and 8 show the regressions for VLDL-C. The was significantly associated with VLDL-C in the younger age s. was Table 7. Regression Coefficients of Selected Variables for VLDL Cholesterol by 10- Year Group In Framingham Men t * < p < t0.001 < p<. *p < f * 1.0* 1.1* 0.9t * 0.3* * 0.11* 0.05* 0.04* 0.02* 0.04*

5 ASSOCIATIONS WITH LIPOPROTEIN CHOLESTEROLS Wilson et al. 277 Table 8. Regression Coefficients of Selected Variables for VLDL Cholesterol by 10- Year Group In Framlngham Women 8.2* 5.0* * 25.5* * 2.9* 3.4* t 0.4* 0.4* * 0.04* 0.03* * * * 3444 * < p < fo.001 < p<. also significantly related to VLDL-C among younger women, but among men the association was only significant for those 30 to 39 years old. The magnitude of the regression coefficient estimates for the men was always greater than that among women for the smoking and variables in the overall VLDL-C analysis. There was only a slight positive association of alcohol intake with VLDL-C in men, and none for women. Table 9 lists the multiple regression coefficients for the number of cigarettes smoked per day from the model that regresses total cholesterol and its lipoprotein cholesterol fractions on alcohol intake,, and cigarette number. In 20- to 49-year-old men and women, cigarette number and HDL-C were strongly associated in an independent fashion, as Table 9. Regression Coefficients for Number of Cigarettes In Regression Model Estimating Llpoproteln Cholesterol Fractions LJpoprotein cholesterol HDL-C LDL-C VLDL-C (20-^9 yrs) Men Women -0.11* -0.25* 0.16t 0.33* 0.12* 0.13* (50-79 yrs) Men Women -0.10t -0.20* * 0.21* * Model: cholesterol = f(alcohol intake,, number of cigarettes). * < p < fo.001 < p<. Table 10. Regression Coefficients of Plasma Chemistry, Hematologic, and Selected Variables for HDL Cholesterol In Framlngham Offspring Years Men Women Variable Current smoking (kg/sq m) (oz/wk) Albumin (g/100 ml) Alkaline phosphatase (ILJ/liter) Urea nitrogen (mg/100 ml) Calcium (mg/dl) Globulin (g/100 ml) Glucose (mg/100 ml) Hematocrit (ml/100 ml) LDH (U/ml) Phosphorus (mg/100 ml) SGOT (U/ml) Uric acid (mg/100 ml) Leucocyte count (1000/ml) Number -2.26* -0.60* 0.53* t 0.15* 2.90* * * -0.75* * -0.57* 0.54* -0.05* 2.56* -0.04* -0.67f -0.79* * -0.51* 0.98* * 0.28* 5.27* f -0.46* * -0.51* 0.97* -2.25* -0.07* 0.28* 5.65* -1.03t -0.41* 0.14 * < p < t0.001 < p<. *p <

6 278 ARTERIOSCLEROSIS VOL 3, No 3, MAY/JUNE 1983 Table 11. Regression Coefficients of Plasma Chemistry, Hematologic, and Selected Variables for LDL Cholesterol In Framlngham Offspring Years Men Women Variable Current smoking (kg/sq m) (oz/wk) Albumin (g/100 ml) Alkaline phosphatase (lu/liter) Urea nitrogen (mg/100 ml) Calcium (mg/dl) Globulin (g/100 ml) Glucose (mg/100 ml) Hematocrit (ml/100 ml) LDH (U/ml) Phosphorus (mg/100 ml) SGOT (U/ml) Uric acid (mg/100 ml) Leucocyte count (1000/ml) Number * < p < < p<. *p < : f * * 0.08t -5.44f : : : 6.16f 0.69* 0.08f -5.38t -0.09* 1.794: * 1.204: : * 0.724: 7.21t 4.82* 0.12* 1.464: 0.07t * t 1.104: : 0.12* 0.734: 6.35t 5.39f 0.12* 1.344: 0.08f 0.90* 0.13 previously reported. 12 The number of cigarettes reported was also associated independently with the other lipoprotein cholesterol fractions. The associations were negative for HDL-C and positive for LDL- C, VLDL-C, and total cholesterol. These data from individuals under 50 years of age are evidence that smoking is accompanied by changes in lipoprotein cholesterols that are potentially atherogenic. As the smoking variable used in Table 9 is the actual number of cigarettes smoked, a dose-response effect is present. A different situation holds in those over 50 years of age. Only HDL-C was significantly associated with cigarette number and the other lipoprotein fractions showed no independent associations when alcohol intake and were also in the model. Table 12. Regression Coefficients of Plasma Chemistry, Hematologic, and Selected Variables for VLDL Cholesterol in Framlngham Offspring Years Variable Current smoking (kg/sq m) (oz/wk) Albumin (g/100 ml) Alkaline phosphatase (lu/liter) Urea nitrogen (mg/100 ml) Calcium (mg/dl) Globulin (g/100 ml) Glucose (mg/100 ml) Hematocrit (ml/100 ml) LDH (U/ml) Phosphorus (mg/100 ml) SGOT (U/ml) Uric acid (mg/100 ml) Leucocyte count (1000/ml) Number * < p < fo.001 < p< : * * 0.154: : 1.234: 0.15 Men 0.664: 0.07* 2.89t 0.154: 0.05* 2.894: 1.51* : t 0.07* -0.34* : * 1.534: 0.44t 0.11 Women 2.13* -3.33* 0.08* -0.33* 0.08* -1.40* 1.68* 0.52* 0.11

7 ASSOCIATIONS WITH LIPOPROTEIN CHOLESTEROLS Wilson et al. 279 and the maintain an inverse association with HDL-C in all s, while alcohol intake has the opposite effect. Obesity is the only variable that shows consistently positive associations in the prediction of LDL-C and VLDL-C for this population. Previously reported results for the Offspring men and women under 50 emphasized the importance of obesity in young adults and its strong association with LDL-C and VLDL-C. 2 These results suggest that obesity is negatively associated with HDL-C both in the younger individuals and those over 50 years, and agrees with the LRC findings. 9 On the other hand, our data suggest that increasing age mitigates the associations of obesity with LDL-C and VLDL-C, and that obesity is less strongly associated with different levels of these lipoprotein cholesterol fractions in those over 50 years. The relationship of these lipoprotein cholesterols to alcohol intake deserves special mention. As in the LRC study, highly significant associations of alcohol consumption and HDL-C were observed. 910 Evidence from the LRC suggests that LDL-C and VLDL- C are also positively related to alcohol intake in teenagers. 10 In our population we found moderate, but significant, negative associations between alcohol intake and LDL-C in young men and women, and generally insignificant associations with VLDL-C. Cigarette smoking is accepted as an important risk factor for cardiovascular disease, but mechanisms to explain the association remain obscure. While certain evidence may favor an effect of short duration on cardiac function rather than more fundamental changes in the atherosclerotic process, the associations demonstrated in this report suggest that the composition of the lipoprotein cholesterol fractions is altered in young adult cigarette smokers. Various effects of smoking on lipids have been reported. They include an increase in total cholesterol, 11 a decrease or no effect in HDL-C, 12> 13 and an increase in LDL-C. 14 ' 15 In our results the positive coefficients for LDL-C and VLDL-C, and negative coefficients for HDL-C with smoking, are of interest. For HDL-C, multiple regression coefficients for smoking in this study are for men and for women. Corresponding values from the LRC investigations for men were for light smokers (1 to 19 cigarettes per day), and -5.3 for heavy smokers (more than 19 cigarettes per day), while the LRC values for women were -4.5 for light smokers and -8.6 for heavy smokers. Similar sorts of deductions can be made using the coefficients in Table 9. For example, a man under 50 years of age who smokes 20 cigarettes a day would be predicted to have an HDL-C 2.2 mg/dl less and a LDL-C 3.2 mg/dl greater than a similar nonsmoker of the same sex. The magnitude of these changes is less than that reported by the LRC, but in the same direction. Furthermore, these analyses demonstrate differences in LDL-C and VLDL-C fractions in smokers over 50 years of age. Taken with previous stud- Tables 10 to 12 display the multiple regression models for the lipoprotein cholesterols regressed on an additional set of independent variables. The results are shown for all the variables considered and separately for those variables that remain significant after a backward stepwise selection process was completed. From the latter analysis, calcium, uric acid, leucocyte count, alkaline phosphatase, smoking, alcohol, and were significantly associated with HDL-C levels for both sexes (Table 10). In Table 11,, calcium, hematocrit, LDH, and leucocyte count were associated with LDL-C for both sexes. Slightly different results were obtained in Table 12. VLDL-C was associated with alkaline phosphatase, glucose, uric acid, and leucocyte count in both sexes. Several other associations between the lipoprotein cholesterols and the clinical chemistry determinations also occurred for each sex separately. Discussion Earlier publications have explored the interrelationships of smoking, obesity, and alcohol intake with lipoprotein cholesterol measurements. The age range and large number of participants in this report allow assessment of effects not studied in previous investigations. The first two tables provide background information. frequency declines for each gender with each succeeding age in our population. Men drink more than twice as much alcohol per week as the women for each age, and alcohol intake peaks between 40 and 59 years of age in both men and women. The data provided by the smoking frequencies, levels, and mean alcohol intake are analyzed further in Tables 3-8. A close examination of Tables 3 and 4 illustrate some of these findings. For instance, in the prediction of HDL-C for the total male and female population, the various coefficients differ for smoking and alcohol intake, but are almost the same for the. These coefficients have direct clinical application. Male smokers have HDL-C levels 3.2 mg/dl less than nonsmoking men. A oneunit increase in the approximates a 5 Ib increase in weight, and is associated with approximately a 0.8 mg/dl decrease in HDL-C. Similarly, a 1 oz/wk increase in alcohol consumption is associated with an HDL-C elevation of 0.4 mg/dl. The age-specific regression analyses yield slightly different results. There are fewer participants in the older age s, and prediction equations are less reliable. The smaller multiple values indicate that the association of the independent variables with the lipoprotein cholesterols are not as strong. Some findings are very consistent, though. The relationship of the independent variables to HDL-C are consistent across age s.

8 280 ARTERIOSCLEROSIS VOL 3, No 3, MAY/JUNE 1983 ies, these results suggest that current smoking may alter lipoprotein cholesterol metabolism; that there is a dose-response effect for the number of cigarettes smoked on HDL-C, LDL-C, and VLDL-C in adults under 50; and that cigarette number may have a greater impact on HDL-C in women than in men. 12 ' 16 The regressions in Tables 10 to 12 with the automated chemistry data complement similar reports of lipoprotein cholesterol levels. 17 Calcium is strikingly associated with higher HDL-C levels, while alkaline phosphatase, uric acid, and leucocyte counts exhibit negative associations. The association between calcium and cholesterol levels has been noted before. 13 These data include individuals whose calcium levels ranged from 7.6 to 11.2 mg/dl. When the same regression was run after stratification by low, middle, and high calcium levels, uniform associations of HDL-C with calcium were still observed. Elevated SGOT levels and low alkaline phosphatase levels have been linked to elevated HDL-C levels In our data, SGOT is not significantly related to HDL-C values, while the association of alkaline phosphatase and HDL-C is negative, as in the LRC results. The pathophysiology underlying these associations remains obscure. Recent data suggest that chronicity and level of alcohol intake are related to HDL-C levels. 19 This sort of information and further data about muscle, liver, and bone metabolism may be needed to sort out these biochemical relationships to HDL-C. Apart from the, serum calcium, LDH, hematocrit, and leucocyte count are also significantly related to LDL-C in men and women. There is no obvious explanation why this set of variables has such an association. The role of other variables is enigmatic; i.e., that of albumin, where a 1 g/liter increase is associated with a 10 mg/dl lower LDL-C level in men, but no change in women. For both men and women, the variables, alkaline phosphatase, glucose, uric acid, and leucocyte count, are independently associated with VLDL-C. Elevations of glucose and uric acid in one study, 20 and glucose, obesity, and prebeta lipoproteins in another 21 have been suggested as covariates that may herald glucose intolerance or diabetes mellitus. The association of the lipoprotein cholesterols and the leucocyte count merits special comment. Greater leucocyte count has been reported as a risk factor for myocardial infarction in an earlier publication. 22 A follow-up investigation suggested that the effect was obtained only in smokers. 23 The data in Tables 10 to 12 further suggest that greater leucocyte counts are associated with a more atherogenic lipid profile: lower HDL-C and greater LDL-C and VLDL-C levels. These associations with the leucocyte count also exist when an analysis is performed on only nonsmokers. Earlier research has hinted that disease states, medication use, and environmental variables may affect levels of lipoprotein cholesterols. This report has highlighted some of these features, emphasizing common environmental variables, but includes less commonly measured factors as well. From a different perspective, the analyses undertaken in this study freshen our thoughts about the potential effects on lipoprotein cholesterol measurements, and reactivates our interest in factors such as liver function, leucocyte count, and calcium levels which might have bearing on the atherosclerotic process. References 1. Castelll WP, Gordon T, Hjortland MC, et al. and blood liplds: The Cooperative Lipoprotein Phenotyplng Study. Lancet 1977;2: Garrison RJ, Wilson PW, Caatelll WP, Felnlelb M, Kannel WB, McNamara P. Obesity and lipoprotein cholesterol in the Framingham Offspring Study. Metabolism 1980;29: Helss G, Johnson NJ, Relland S, et al. The epidemiology of plasma high-density lipoprotein cholesterol levels. The LJpid Research Clinics Program Prevalence Study. Summary. Circulation 1980;62(Suppl (V):IV-116-IV LlpkJ Research Clinics Program. Manual of Laboratory Operation, vol. 1: LJpid and lipoprotein analysis. DHEW publication no. (NIH) Bethesda, Maryland: National Institutes of Health 5. Hjortland MC, Felnlelb M. Epidemlologjc relationships of alcohol intake and coronary heart disease: results from NHLBI programs. In: Avogaro P, Sirtori CR, Tremoli E, eds. Metabolic effects of alcohol. New York: Elsevier/North Holland Biomedical Press, 1979: Aboil LL, Levy BB, Brodle BB, Kendall FE. A simplified method for the estimation of total cholesterol in serum and demonstration of its specificity. J Biol Chem 1952;195: Kessler G, Lederer H. Fluorometric measurements of triglycerfdes. In: Skegg LT Jr. ed. Technicon symposium: Automation In analytical chemistry, New York: Medical Inc, 1966: Neter J, Wasserman W. Applied linear statistical models. Illinois: Richard D. Irwin Press, 1974: Glueck CJ, Taylor HL, Jacobs D, Morrison JA, Beaglehole R, Williams OD. Plasma high-density lipoprotein cholesterol: Association with measurements of body mass. Circulation 1980;62 (Suppl IV):IV-62-IV Glueck CJ, Helss G, Morrison JA, Khoury P, Moore M. intake, cigarette smoking and plasma llpkjs and lipoproteins in year old children. The Collaborative LJpid Research Clinics Prevalence Study. Circulation 1981;64 (Suppl lll):lll-48-lll McGIII HC Jr. Potential mechanisms for the augmentation of atherosclerosis and atherosclerotic disease by cigarette smoking. Prev Med 1979;8: Garrison RJ, Kannel WB, Felnlelb M, Castelll WP, McNamara PM, Padgett SJ. Cigarette smoking and HDL cholesterol: The Framingham Offspring Study. Atherosclerosis 1978;30: Phillips NR, Havel RJ, Kane JP. Levels and interrelationships of serum and lipoprotein cholesterol and triglycerides. Association with adiposity and the consumption of ethanol, tobacco, and beverages containing caffeine. Arteriosclerosis 1981;1: Heyden S, Helss G, Manegold C, et al. The combined effect of smoking and coffee drinking on LDL and HDL cholesterol. Circulation 1979:60: Garrison RJ, Castelll WP, Felnlelb M, et al. The association

9

10 Factors associated with lipoprotein cholesterol levels. The Framingham study. P W Wilson, R J Garrison, R D Abbott and W P Castelli Arterioscler Thromb Vasc Biol. 1983;3: doi: /01.ATV Arteriosclerosis, Thrombosis, and Vascular Biology is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX Copyright 1983 American Heart Association, Inc. rights reserved. Print ISSN: Online ISSN: The online version of this article, along with updated information and services, is located on the World Wide Web at: published Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally in Arteriosclerosis, Thrombosis, and Vascular Biology can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: Subscriptions: Information about subscribing to Arteriosclerosis, Thrombosis, and Vascular Biology is online at: