Human consumption of food usually is evenly distributed. Epidemiology

Transcription

1 Epidemiology Fasting and Nonfasting Lipid Levels Influence of Normal Food Intake on Lipids, Lipoproteins, Apolipoproteins, and Cardiovascular Risk Prediction Anne Langsted, BMSc; Jacob J. Freiberg, MD; Børge G. Nordestgaard, MD, DMSc Background Lipid profiles are usually measured after fasting. We tested the hypotheses that these levels change only minimally in response to normal food intake and that nonfasting levels predict cardiovascular events. Methods and Results We cross-sectionally studied individuals 2 to 95 years of age from the Copenhagen General Population Study. We also studied 9319 individuals 2 to 93 years of age from the Copenhagen City Heart Study, 1166 of whom developed cardiovascular events during 14 years of follow-up. Compared with fasting levels, total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein (HDL) cholesterol, and albumin levels were reduced up to 3 to 5 hours after the last meal; triglycerides levels were increased up to 6 hours after the last meal; and non-hdl cholesterol level, apolipoprotein A1 level, apolipoprotein B level, ratio of total cholesterol to HDL cholesterol, and ratio of apolipoprotein B to apolipoprotein A1 did not change in response to normal food intake. The maximum changes after normal food and fluid intake from fasting levels were.2 mmol/l for total cholesterol,.2 mmol/l for low-density lipoprotein cholesterol,.1 mmol/l for HDL cholesterol, and.3 mmol/l for triglycerides. Highest versus lowest tertile of nonfasting total cholesterol, non-hdl cholesterol, low-density lipoprotein cholesterol, apolipoprotein B, triglycerides, ratio of total cholesterol to HDL cholesterol, and ratio of apolipoprotein B/apolipoprotein A1 and lowest versus highest tertile of nonfasting HDL cholesterol and apolipoprotein A1 predicted 1.7- to 2.4-fold increased risk of cardiovascular events. Conclusions Lipid profiles at most change minimally in response to normal food intake in individuals in the general population. Furthermore, nonfasting lipid profiles predicted increased risk of cardiovascular events. (Circulation. 28; 118: ) Key Words: apolipoproteins cardiovascular diseases lipids lipoproteins Human consumption of food usually is evenly distributed throughout the day (ie, in the form of 3 meals and snacks in between). The fasting state occurs, by definition, after an 8-hour fast 1 ; thus, most humans find themselves in the nonfasting state for the majority of a 24-hour period, perhaps with the exception of the early morning hours. Clinical Perspective p 256 Despite this fact, plasma lipids, lipoproteins, and apolipoproteins for cardiovascular risk prediction are usually measured in the fasting state. 1 3 A main reason is the increase in triglyceride levels seen during a fat tolerance test, in which patients typically consume 1 g fat per 1 kg body weight. 4,5 However, levels of nonfasting triglycerides are better at predicting future cardiovascular events than levels of fasting triglycerides. 6,7 Furthermore, it is possible that nonfasting levels of lipids, lipoproteins, and apolipoproteins differ only minimally from levels in the fasting state simply because most people consume far less fat at ordinary meals than during a fat tolerance test. We tested the hypothesis that levels of total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, non-hdl cholesterol, triglycerides, apolipoprotein B, and apolipoprotein A1 and the ratios of total cholesterol to HDL cholesterol and apolipoprotein B to apolipoprotein A1 change only minimally in response to normal food intake in individuals in the general population. For this purpose, we cross-sectionally examined individuals from the Copenhagen General Population Study. We also tested whether nonfasting levels of these lipids, lipoproteins, and apolipoproteins predict increased risk of cardiovascular events. To test these hypotheses, we used Received February 6, 28; accepted August 21, 28. From the Department of Clinical Biochemistry and the Copenhagen General Population Study, Herlev Hospital, Copenhagen University Hospital, Herlev (A.L., J.J.F., B.G.N.), and Copenhagen City Heart Study, Bispebjerg Hospital, Copenhagen University Hospital, University of Copenhagen, Copenhagen (B.G.N.), Denmark. The online-only Data Supplement can be found with this article at Correspondence to Børge G. Nordestgaard, Professor, Chief Physician, Department of Clinical Biochemistry, Herlev Hospital, Copenhagen University Hospital, Herlev Ringvej 75, DK-273 Herlev, Denmark. brno@heh.regionh.dk 28 American Heart Association, Inc. Circulation is available at DOI: /CIRCULATIONAHA

2 248 Circulation November 11, individuals from the Copenhagen City Heart Study whom we followed up for 14 years, during which time 1166 experienced cardiovascular events. Methods Participants Participants gave written informed consent. The studies were approved by the Herlev Hospital and by the Danish ethical committee for Copenhagen and Frederiksberg (No /91 and 1-144/ 1). The participants in the 2 studies filled out the same questionnaire and were examined similarly; however, information on time since the last meal was obtained only in the cross-sectional study and on different individuals, not on the same individuals several times. Cross-Sectional Study On the basis of the Danish Central Person Registration number, the Copenhagen General Population Study (23 to 26 examination) and Copenhagen City Heart Study (21 to 23 examination) 7 recruited participants randomly from the general population of Copenhagen, Denmark; because 81% of all participants were from the Copenhagen General Population study, we refer only to this study for the cross-sectional data. Importantly, lipid levels were measured in the same laboratory for both studies, and the collection of all other data was identical for the 2 studies. From 21 through 26, we examined men and women 2 to 95 years of age, 142 of whom were excluded as outliers (ie, participants with fasting or nonfasting levels of total cholesterol, LDL cholesterol, HDL cholesterol, non-hdl cholesterol, triglycerides, apolipoprotein A1, apolipoprotein B, or albumin outside 3 SD from the mean). We excluded outliers because chance clustering of outliers in 1 group versus another could bias the results. This left individuals for the present cross-sectional analyses. The examination included a self-administered questionnaire on detailed medical history and lifestyle, which was checked by an examiner on the day of attendance. In addition, a physical examination was performed, and blood samples were drawn for measurements of total cholesterol, LDL cholesterol, HDL cholesterol, non-hdl cholesterol, triglycerides, apolipoprotein A1, apolipoprotein B, and albumin. Before blood sampling, participants were asked when they had eaten their last meal ( to 1, 1 to 2, 2 to 3, 3 to 4, 4 to 5, 5 to 6, 6 to 7, 7 to 8, or 8 hours ago [fasting]); this information was obtained to test the hypothesis of the present study. The time of day for blood sampling also was registered (between 8 AM and 7 PM). Prospective Study The Copenhagen City Heart Study is a prospective cardiovascular study of the Danish general population. 7 At the third examination in 1991 through 1994, participated. In the present study, we included 9319 individuals from this examination who had nonfasting total cholesterol, LDL cholesterol, HDL cholesterol, non-hdl cholesterol, triglycerides, apolipoprotein B, and apolipoprotein A1 determined and who were free of ischemic cardiovascular disease at baseline. End points used in the study were fatal and nonfatal ischemic cardiovascular events, including myocardial infarction and ischemic stroke. 7,8 Participants were followed up through the use of their unique Central Person Registration number from baseline, defined as the date of participation in the 1991 through 1994 examination, and until the occurrence of an ischemic cardiovascular event, death, or July 27, whichever came first. Follow-up was 1% complete. Information on diagnosis of ischemic cardiovascular events was collected by reviewing all hospital admissions and diagnoses entered in the Danish National Hospital Discharge Register and all causes of death entered in the Danish National Register of Causes of Death and was verified by medical records from hospitals and general practitioners. 7,8 Table 1. Characteristics of Individuals From the General Population Characteristics Cross-Sectional Study, CGPS Prospective Study, CCHS N Women, % Age, y 6 (5 69) 59 (46 7) Total cholesterol, mmol/l 5.6 ( ) 6. ( ) Non-HDL 4. ( ) 4.4 ( ) cholesterol, mmol/l LDL cholesterol, mmol/l 3.2 ( ) 3.7 ( ) HDL cholesterol, mmol/l 1.6 (1.3 2.) ( ) Triglycerides, mmol/l 1.4 (1. 2.1) ( ) Apolipoprotein A1, mg/dl 154 ( ) 139 (122 16) Apolipoprotein B, mg/dl 19 (9 131) 84 (7 1) Total cholesterol/hdl ( ) 3.9 ( ) cholesterol Apolipoprotein.7 (.6.9).6 (.5.8) B/apolipoprotein A1 Albumin, mol/l 615 ( ) Body mass index, kg/m 2 26 (23 28) 25 (22 28) Alcohol intake, g/d 14 (5 26) 1 (3 2) Hypertension, % Physical inactivity, % 8 12 Smokers, % Diabetes mellitus, % 5 4 Lipid-lowering therapy, % 8.5 Postmenopausal, women only, % Postmenopausal with HRT, women only, % CGPS indicates Copenhagen General Population Study; CCHS, Copenhagen City Heart Study, 1991 to 1994 examination; and HRT, hormone replacement therapy. Continuous variables are shown as median (interquartile range). Risk Factors Men consuming 36 g of alcohol per day ( 21 U/wk) and women consuming 24 g/d ( 14 U/wk) were defined as heavy alcohol drinkers. Hypertension was defined as systolic blood pressure 14 mm Hg, diastolic blood pressure 9 mm Hg, or use of antihypertensive medication. Smokers were active smokers. Physical inactivity was leisure-time activity of 4 h/wk and predominantly sedentary work. Diabetes mellitus was self-reported disease, use of insulin or oral hypoglycemic pills, or a nonfasting plasma glucose 11 mmol/l. Participants reported lipid-lowering therapy, and women reported menopausal status and use of hormone replacement therapy. Analysis Fresh blood samples were analyzed with standard hospital assays to measure plasma levels of total cholesterol, triglycerides, LDL cholesterol, HDL cholesterol, apolipoprotein A1, apolipoprotein B, and albumin (Konelab, Helsinki, Finland or Boehringer Mannheim, Mannheim, Germany) (Table I of the online-only Data Supplement); coefficients of variations varied from 2% to 4%, and drift over time from year to year varied from 1% to 4%. Assays were followed up daily for precision and 4 to 12 times yearly for accuracy with a Scandinavian quality control program. LDL cholesterol was calculated from the Friedewald equation if triglycerides were 4 mmol/l 1

3 Langsted et al Normal Food Intake, Lipid Profiles, and CVD Cholesterol (mmol/l) *** ** * *** *** *** Total cholesterol Non-HDL cholesterol ** LDL cholesterol ** *** *** ** * HDL cholesterol Fasting Time since last meal (hours) N= and measured directly at higher triglyceride levels. Non-HDL cholesterol was total cholesterol minus HDL cholesterol. Statistical Analysis We used Stata 9.2 (Stata Corp, College Station, Tex). Plasma triglycerides and HDL cholesterol were skewed and hence were transformed logarithmically to approach normal distributions. In the cross-sectional study, because age, sex, and other covariates differed as a function of time since the last meal (online-only Data Supplement Table II) and because age, sex, and other covariates influence lipid levels, general linear models were used to adjust for age and sex; for age, sex, and albumin; or for all covariates listed in online-only Data Supplement Table II. For the adjusted analyses, men and women were stratified into 5-year age groups from 2 to 8 years of age, and albumin values were stratified according to time since the last meal (or time of sampling). Differences in levels of lipids, lipoproteins, apolipoproteins, and albumin as a function of time since the last meal ( to 1, 1 to 2, 2 to 3, 3 to 4, 4 to 5, 5 to 6, 6to7,7to8,or 8 hours) were tested with the Student t test between fasting levels ( 8 hours) and the 8 other time points. Differences in these levels also were tested according to time of day of blood sampling, with levels sampled between 8 and 9 AM as reference. All tests were corrected for multiple comparisons with the Bonferroni method (P values were multiplied by the number of parallel tests). 9 In the prospective study, Cox proportional-hazards regression models with age, blood pressure, smoking status, use of lipidlowering drugs, and use of hormone therapy as covariates or with age, blood pressure, smoking status, use of lipid-lowering drugs, use of hormone therapy, body mass index, diabetes mellitus, and high-sensitivity C-reactive protein levels as covariates were used to estimate hazard ratios for ischemic cardiovascular events by tertiles of total cholesterol, non-hdl cholesterol, LDL cholesterol, HDL Apolipoprotein (mg/dl) Albumin (micromol/l) *** *** *** *** *** *** *** *** Apolipoprotein A1 Apolipoprotein B *** *** Triglycerides Albumin Total cholesterol/hdl cholesterol Apolipoprotein B/Apolipoprotein A Fasting Time since last meal (hours) Figure 1. Levels of lipids, lipoproteins, apolipoproteins, and albumin as a function of time since the last meal. Values are means with SE adjusted for sex and age. With adjusted values, Bonferroni-corrected P values on unpaired Student t test vs fasting levels ( 8 hours since the last meal) are as follows: *P.5, **P.1, and ***P.1. cholesterol, apolipoprotein A1, apolipoprotein B, triglycerides, ratio of apolipoprotein B to apolipoprotein A1, and ratio of total cholesterol to HDL cholesterol. Hazard ratios were corrected for regression dilution bias with a nonparametric method. 7,1 We used the nonparametric method because it does not rely on assumptions of constant variances. For this correction, we used nonfasting values from 4441 individuals not on lipid-lowering therapy who attended both the baseline 1991 to 1994 examination and the 21 to 23 examination; however, the main analyses were conducted on 9319 individuals. The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written. Results Characteristics of individuals from the Danish general population, the Copenhagen General Population Study and the Copenhagen City Heart Study, are shown in Table 1; 1% and 99% of the participants were white and of Danish descent in the 2 studies, respectively. Five percent were on lipidlowering therapy in the cross-sectional study; when these 5% were excluded from the analyses, the results were similar to those presented below. Time Since the Last Meal After adjustment for age and sex and compared with fasting levels, total cholesterol, LDL cholesterol, HDL cholesterol, and albumin were reduced for up to 3 to 5 hours after the last meal, whereas triglycerides increased for up to 6 hours after the last meal (Figure 1). Levels of non-hdl cholesterol, Triglycerides (mmol/l) Ratio

4 25 Circulation November 11, 28 Cholesterol (mmol/l) N= * Total cholesterol Non-HDL cholesterol * * LDL cholesterol HDL cholesterol Fasting Time since last meal (hours) apolipoprotein A1, and apolipoprotein B and ratios of total to HDL cholesterol and apolipoprotein B to A1 did not change in response to normal food intake. After further adjustment for all covariates shown in online-only Data Supplement Table II, the results were similar to those shown in Figure 1. In addition, if we based the analyses in Figure 1 on all individuals without excluding outliers, the results were similar (online-only Data Supplement Figure I). Reduced levels of albumin after normal food intake are likely caused by hemodilution resulting from fluid intake; therefore, we also corrected all values in Figure 1 for mean changes in albumin levels as a function of time since the last meal (Figure 2). After this adjustment, total cholesterol, LDL cholesterol, and albumin levels no longer differed as a function of time since the last meal. However, plasma triglycerides and apolipoprotein B were increased and HDL cholesterol was decreased for up 4 to 6 hours after normal food intake even after this adjustment. Time of Day for Blood Sampling In individuals in the general population with normal food intake, levels of total cholesterol, LDL cholesterol, HDL cholesterol, non-hdl cholesterol, apolipoprotein A1, and apolipoprotein B did not differ substantially, depending on what time of day blood was drawn (Figure 3). However, triglyceride levels were higher between noon and 5 PM compared with earlier or later in the day. Albumin levels were higher later in the day compared with between 8 and 9 AM, Apolipoprotein (mg/dl) Albumin (micromol/l) ** ** ** Apolipoprotein A1 Apolipoprotein B *** *** *** *** *** Triglycerides Albumin Fasting Time since last meal (hours) *** Total cholesterol/hdl cholesterol Apolipoprotein B/Apolipoprotein A1 Figure 2. Levels of lipids, lipoproteins, apolipoproteins, and albumin as a function of time since the last meal. Values are means with SE adjusted for sex, age, and albumin levels. With adjusted values, Bonferroni-corrected P values on unpaired Student t test vs fasting levels ( 8 hours since the last meal) are as follows: *P.5, **P.1, and ***P.1. possibly because of hemoconcentration. The rise in triglyceride levels from noon through 5 PM did not disappear after adjustment for albumin levels (data not shown). The results shown in Figure 3 were adjusted for age and sex; however, even after further adjustment for all covariates shown in Table II of the online-only Data Supplement, results were similar. Time Since the Last Meal Stratified for Time of Day Levels of total cholesterol and LDL cholesterol did not change in response to normal food intake when blood samples were drawn in the morning or afternoon, whereas these 2 variables decreased for up to 5 hours after normal food intake compared with fasting levels when blood samples were drawn in the evening (data not shown). Plasma triglycerides increased in response to normal food intake regardless of whether blood was drawn in the morning, afternoon, or evening (data not shown). Maximum Change in Lipids, Lipoproteins, and Apolipoproteins After normal food intake, individuals in the general population had a maximum mean change from fasting levels of.2 mmol/l for total cholesterol at to 2 hours after the last meal,.2 mmol/l for LDL cholesterol at to 2 hours,.1 mmol/l for HDL cholesterol at to 5 hours, and.3 mmol/l for triglycerides at 1 to 4 hours after the last meal Ratio Triglycerides (mmol/l)

5 Langsted et al Normal Food Intake, Lipid Profiles, and CVD Cholesterol (mmol/l) am 9-1 am 1-11 am am 12-1 pm Total cholesterol Non-HDL cholesterol LDL cholesterol * HDL cholesterol 1-2 pm 2-3 pm 3-4 pm 4-5 pm 5-6 pm 6-7 pm Time of day for blood sampling N= (Table 2 and Figure 1). These results were adjusted for age and sex; however, even after further adjustment for all covariates shown in online-only Data Supplement Table II, results were similar. Risk of Cardiovascular Events In the Copenhagen City Heart Study, we observed 1166 cardiovascular events (627 myocardial infarction events and 539 ischemic stroke events) during 14 years of follow-up. Highest versus lowest tertile of nonfasting total cholesterol, non-hdl cholesterol, LDL cholesterol, apolipoprotein B, triglycerides, ratio of total to HDL cholesterol, and ratio of apolipoprotein B to A1 and lowest versus highest tertile of nonfasting HDL cholesterol and apolipoprotein A1 predicted a 1.7- to 2.4-fold increase in risk of cardiovascular events (Table 3). Discussion We found that levels of lipids, lipoproteins, and apolipoproteins at most changed minimally in response to normal food intake in individuals in the general population. Furthermore, nonfasting levels of these lipids, lipoproteins, and apolipoproteins all predicted increased risk of cardiovascular events. To the best of our knowledge, no substantial evidence demonstrates that fasting lipid levels are superior to nonfasting levels for cardiovascular risk prediction. It is therefore reasonable to review the arguments often used in favor of fasting versus nonfasting lipid measurements, simply because Apolipoprotein (mg/dl) Albumin (micromol/l) am 9-1 am 1-11 am am 12-1 pm Apolipoprotein A1 Apolipoprotein B *** ** *** * * Triglycerides *** *** *** *** *** *** *** *** *** *** Albumin Total cholesterol/hdl cholesterol Apolipoprotein B/Apolipoprotein A1 1-2 pm 2-3 pm 3-4 pm 4-5 pm 5-6 pm 6-7 pm Time of day for blood sampling Figure 3. Levels of lipids, lipoproteins, apolipoproteins, and albumin as a function of time of day for blood sampling. Values are means with SE adjusted for sex and age. With adjusted values, Bonferroni-corrected P values on unpaired Student t test vs morning levels (at 8to9AM) are as follows: *P.5, **P.1, and ***P.1. * the fasting requirement possibly makes blood sampling unnecessarily difficult for millions of patients worldwide. One argument often presented in favor of measuring lipids, lipoproteins, and apolipoproteins in the fasting state is the increase in triglyceride levels seen during fat tolerance tests. 4,5,7 However, our findings show that levels of lipids, lipoproteins, and apolipoproteins after normal food intake differ only minimally from levels in the fasting state, probably because most people consume much less fat at ordinary meals than during a fat tolerance test. Another argument often presented is that calculating LDL cholesterol with the Friedewald equation requires fasting triglyceride measurement. 1 However, our data demonstrate that calculated LDL cholesterol changes only minimally in response to normal food intake and that increased nonfasting LDL cholesterol predicts increased risk of cardiovascular events. Therefore, it is tempting to speculate that the main reasons for measuring lipid levels in the fasting rather than the nonfasting state are simply that it has become the norm worldwide and that the fasting requirement has been applied in almost all randomized lipid-lowering trials. Total and LDL cholesterol levels were reduced for up to 3 to 4 hours after normal food intake; this reduction was paralleled by a similar fall in albumin levels. Therefore, and because total and LDL cholesterol levels did not change in response to normal food intake after adjustment for albumin levels, the explanation for the small reduction in total and Ratio Triglycerides (mmol/l)

6 252 Circulation November 11, 28 Table 2. Change in Lipids, Lipoproteins, and Apolipoproteins Compared With Fasting as a Function of Time Since the Last Meal in the General Population (Copenhagen General Population Study) Time Since Last Meal, h Characteristics n Total cholesterol, mmol/l.2 (.2.1).2 (.2.1).1 (.2.)*.1 (.1.) LDL cholesterol, mmol/l.2 (.3.1).2 (.2.1).1 (.2.1).1 (.2.) HDL cholesterol, mmol/l.1 (.1.).1 (.1.).1 (.1.).1 (.1.) Non-HDL cholesterol, mmol/l.1 (.2.).1 (.1.). (.1.). (.1.1) Triglycerides, mmol/l.2 (.2.3).3 (.2.3).3 (.2.3).3 (.2.4) Apolipoprotein A1, mg/dl 2.6 ( 4.7.5) 2.6 ( 4.7.6) 2. ( 4..1) 1. ( ) Apolipoprotein B, mg/dl.5 ( ).6 ( 2.8) 1.2 (.9 3.4) 2.6 (.4 4.8) Total cholesterol/hdl cholesterol. (.1.1).1 (..2).1 (..2).1 (..2) Apolipoprotein B/apolipoprotein A1. (.1.1). (.1.1). (.1.1). (.1.1) Albumin, mol/l 19 ( 23 15) 19 ( 23 15) 15 ( 19 11) 1 ( 14 6) (Continued) Values are mean change (95% confidence interval). Bonferroni corrected P values on Student t test (unpaired) vs fasting levels ( 8 hours since last meal): *P.5, P.1, and P.1. LDL cholesterol levels is most likely hemodilution resulting from fluid intake in relation to the meal. A similar fall in LDL cholesterol, but not in total cholesterol, from fasting levels was observed among 115 subjects 3 to 5 hours after a normal breakfast. 11 Interestingly, the National Cholesterol Education Program 2 advises that total cholesterol, but not LDL cholesterol, be used in the nonfasting state, whereas our data suggest that the change in the 2 after normal food intake is similar but minimal. Triglycerides increased and HDL cholesterol decreased in response to normal food intake even after the correction for albumin levels and thus correction for hemodilution due to fluid intake. Therefore, the changes in these levels are most likely due to food intake rather than fluid intake. Although the increases in triglycerides are likely attributable directly to fat intake, the parallel reduction in HDL cholesterol likely is due to bidirectional lipid exchange between triglyceride-rich lipoproteins and HDL particles: Lipid transfer proteins mediate transfer of triglycerides from triglyceride-rich lipoproteins to HDL, with back-transfer of cholesteryl ester from HDL to triglyceride-rich lipoproteins. 12 In accordance with our study, triglycerides increased and HDL cholesterol decreased among 115 subjects 3 to 5 hours after a normal breakfast compared with fasting levels. 11 The National Cholesterol Education Program 2 advises that HDL cholesterol be used in the nonfasting state. The only modest increase in triglyceride levels during normal food intake, together with the recent demonstration of high predictive ability of nonfasting triglycerides for risk of cardiovascular events 6 and risk of myocardial infarction, ischemic heart disease, and death, 7 suggests the possibility that nonfasting rather than fasting triglyceride levels could be used for cardiovascular risk prediction. Because we show here that other lipids, lipoproteins, and apolipoproteins, despite being measured in the nonfasting state, all predicted increased risk of cardiovascular events, perhaps nonfasting levels of all lipids, lipoproteins, and apolipoprotein could be used for cardiovascular risk prediction. If implemented, this would simplify blood sampling for lipid profile measurements. It should be considered whether maximum increases in triglycerides of.3 mmol/l, maximum decreases in total and LDL cholesterol of.2 mmol/l, and maximum decreases in HDL cholesterol of.1 mmol/l in response to normal food and fluid intake are of clinical importance. First, analytical and other physiological variation also may influence levels of lipids, lipoproteins, and apolipoproteins, 1 and this variation could be as large as or even larger than that attributed to normal food intake in the present study. Second, cardiovascular risk assessment is not based only on levels of lipids and lipoproteins, but these values are used with several other risk factors to estimate absolute 1-year risk of ischemic heart disease or cardiovascular death. 2,3 Therefore, these minor changes in lipid and lipoprotein levels after normal food intake will have only minimal influence on the estimated 1-year risk and thus on whether patients should be offered lipid-lowering therapy. Third, neither American nor European guidelines on cardiovascular disease prevention use triglycerides for calculating 1-year risk, 2,3 the lipid value mostly affected by normal food intake; American guidelines suggest evaluating non-hdl cholesterol rather than triglycerides, 2 and non-hdl cholesterol was unaffected by normal food intake in the present study. Fourth, because European guidelines suggest that cardiovascular risk prediction is just as good regardless of whether HDL cholesterol is included, a minimal change in HDL cholesterol of.1 mmol/l resulting from normal food intake certainly would not have a major impact on cardiovascular risk prediction. Finally, statin trials have documented that clinically important reductions in total and LDL cholesterol should be 1 mmol/l, 2,3 also questioning whether a change in the detected levels of total and LDL cholesterol of.2 mmol/l resulting from normal food intake is clinically important. Most previous studies have focused on the change in levels of lipids, lipoproteins, and apolipoproteins after a fat toler-

7 Langsted et al Normal Food Intake, Lipid Profiles, and CVD 253 Table 2. Continued Time Since Last Meal, h Fasting (.1.). (.1.1).1 (..2). (.1.2)..1 (.2.). (.1.1).1 (..2).1 (..3)..1(.1.)*.1 (.1.). (.1.).1 (.1.).. (.1.1). (.1.1).1 (..3).1 (.1.2)..2 (.2.3).1 (.1.2).1 (..2). (.1.1)..9 ( ).8 ( 3.1 ).7 ( 3.8 ) 3.4 ( 7..2). 1.1 ( ) 1.8 (.8 4.3) 2.9 (.6 6.4) ( )..1 (..2).1 (..2).1 (..3).1 (..3).. (.1.1). (.1.1). (.1.2). (.2.2). 2( 6 3) 4 ( 1 8) 4 ( 2 1) 2( 1 6). ance test rather than after normal food intake. These studies usually have the participants consume a meal containing a so-called oral fat load of 1 g of fat per 1 kg body weight 5,7,13,14 and detect increases in triglycerides of 1 to 2 mmol/l. However, most studies find that 3 g fat in a meal has no or very little effect on postprandial lipidemia, which is in accordance with the present demonstration of at most minimal changes in levels of lipids, lipoproteins, and apolipoproteins in response to normal food intake in individuals in the general population. In support of our findings that nonfasting lipids, lipoproteins, and apolipoproteins predict cardiovascular events, previous studies including 7735 men and 258 men found that upper versus lower quartiles or quintiles of nonfasting levels of total cholesterol, LDL cholesterol, triglycerides, and apolipoprotein B predicted a 2- to 3-fold increased risk of ischemic heart disease. 18,19 This is similar to the corresponding risk estimates found for upper versus lower tertiles or quintiles of mainly fasting levels of LDL cholesterol, triglycerides, and apolipoprotein B of 2- to 3-fold Similarly, lower versus upper quartiles or quintiles of nonfasting HDL cholesterol and apolipoprotein A1 predicted a 2- to -fold risk of ischemic heart disease in men, 18,19 whereas corresponding values for mainly fasting levels were 2- to 3-fold. 2,22 Importantly, nonfasting triglycerides may be as good as or even better than fasting triglycerides at predicting risk of cardiovascular events. 6,7,23 In addition, apolipoproteins can be used as predictors measured in the nonfasting state, 24 as confirmed in the present study. One limitation of the present study is that we did not draw blood from the same individuals in the fasting state and at fixed intervals after the last meal. However, even after adjustment for all covariates shown in Table II of the online-only Data Supplement, the results were similar to those present in our study. Nevertheless, significant differences can be seen in clinical and demographic characteristics in fasting versus nonfasting participants, as illustrated in Table II of the online-only Data Supplement, and although multivariable adjustment was applied in an attempt to address these differences, the concern about residual confounding in a nonrandomized observational study cannot be totally excluded. The relatively small number of participants with fasting status as the comparison group is another limitation because it might decrease power to detect statistically significant differences between fasting and nonfasting values. A third limitation is that we had no record of exactly what each participant had eaten before blood sampling. On the other hand, our study represents the real world where usually no such information is available before blood sampling. Fourth, it could be argued that the time since the last meal was self-reported and subject to error; however, the time since the last meal was not part of the questionnaire but was asked directly by the examiner at the exact time of blood sampling. Fifth, one could argue that fasting before lipid profile measurement is done to improve the accuracy of LDL cholesterol measurements calculated partly from triglycerides and that we demonstrated an increase in triglycerides in response to normal food intake. However, despite the increase in triglycerides, the calculated LDL cholesterol levels did not change after normal food intake after adjustment for hemodilution resulting from fluid intake. Sixth, the participants in the cross-sectional study were recruited in 21 to 26, and the participants in the prospective study were recruited in 1991 to 1994 and then followed up for 14 years, which imply some differences in levels of cardiovascular risk factors in the 2 studies. Such differences could affect the results; however, in the prospective study, we adjusted hazard ratios for conventional cardiovascular risk factors. Finally, in the present study, we include mainly whites of Danish descent ( 99%); therefore, our results may not necessarily apply to other ethnic groups. Future research could include studies on the predictive value of levels of lipids, lipoproteins, and apolipoproteins for cardiovascular disease risk according to the time since the last meal before blood sampling. Interestingly, a recent metaanalysis found that fasting and nonfasting triglyceride levels were equally good at predicting increased risk of coronary heart disease, 21 whereas another recent study showed that

8 254 Circulation November 11, 28 Table 3. Risk of Cardiovascular Events as a Function of Nonfasting Lipids, Lipoproteins, and Apolipoproteins in the General Population (Copenhagen City Heart Study) Women Approximate Tertiles Incidence/1 person-y Hazard Ratio (95% CI) n Model 1 Model 2 P for Trend Total cholesterol, mmol/l Lower tertile Middle tertile (.8 1.8) 1.4 (.9 2.3) Higher tertile ( ) 1.9 ( ) Non-HDL cholesterol, mmol/l Lower tertile Middle tertile (1.-2.4) (1. 2.4) Higher tertile ( 3.4) 2.1 ( ) LDL cholesterol, mmol/l Lower tertile Middle tertile (.9 2.1) 1.6 (1. 2.4) Higher tertile ( ) 2.2 ( ) HDL cholesterol, mmol/l Higher tertile Middle tertile (1. 1.8) 1.2 (.9 1.7) Lower tertile ( ) 1.7 ( ) Apolipoprotein A1, mg/dl Higher tertile Middle tertile (.9 1.8) 1.2 (.8 1.7) Lower tertile ( ) 1.6 ( ) Apolipoprotein B, mg/dl Lower tertile Middle tertile (1. 1.8) 1.4 (1. 1.9) Higher tertile ( ) 1.8 ( ) Triglycerides, mmol/l Lower tertile Middle tertile (.7 1.6).9 (.6 1.4) Higher tertile ( ) 1.4 (.9 2.2) Total cholesterol/hdl cholesterol Lower tertile Middle tertile (1.3 ) 1.8 ( ) Higher tertile ( ) 2.4 ( ) Apolipoprotein B/apolipoprotein A1 Lower tertile Middle tertile ( ) 1.3 (1. 1.8) Higher tertile ( ) 2.1 ( 2.8) (Continued) CI indicates confidence interval. Model 1 was adjusted for age, blood pressure, smoking, use of lipid-lowering drugs, and use of hormone replacement therapy. Model 2 was adjusted for the covariates in model 1 plus diabetes mellitus, body mass index, and high-sensitivity C-reactive protein. Model 2 has slightly fewer participants than model 1 because of the reduced availability of covariates. All hazard ratios were adjusted for regression dilution bias. nonfasting triglycerides were better than fasting triglycerides at predicting cardiovascular events. 6 Furthermore, we recently documented that a nonfasting triglyceride level of 5 versus 1 mmol/l predicts an age-adjusted 5- and 17-fold increased risk of myocardial infarction in men and women. 7 Therefore, and because atherogenesis may be a postprandial phenomenon, 15,25,26 future research should focus on studies reducing the levels of nonfasting triglycerides and thus remnant lipoprotein cholesterol in an attempt to reduce the risk of cardiovascular disease and death further than that currently obtained by reducing mainly LDL cholesterol levels.

9 Langsted et al Normal Food Intake, Lipid Profiles, and CVD 255 Table 3. Continued Men Approximate Tertiles n Incidence/1 person-y Hazard Ratio (95% CI) Model 1 Model 2 P for Trend (.6 1.4) 1.1 (.7 1.6) ( ) 1.7 (1.1 ) (.6 1.4) 1.1 (.7 1.8) ( ) 2.2 ( ) (.8 1.7) 1.3 (.8 2.) ( ) 1.8 ( ) (1.1 2.) 1.6 ( ) (1.4 ) 2. ( ) (1. 1.9) 1.2 (.9 1.6) (1.3 ) 1.7 ( ) (.8 1.4) 1.1 (.8 1.6) ( ) 1.8 ( ) (.9 2.1) 1.3 (.8 2.1) ( ) 1.8 ( ) (.8 ) 1.1 (.8 1.6) ( 2.8) 2.2 ( ) (.8 1.4) 1.1 (.8 ) ( ) 2.1 ( 2.8) If nonfasting rather than fasting lipid profiles for cardiovascular risk prediction were used, it would simplify clinical care for patients worldwide. Because we detected only minimal changes in levels of lipids, lipoproteins, and apolipoproteins in response to normal food intake in the general population, changes that are clinically unimportant, and because nonfasting levels predict cardiovascular events, our data challenge the necessity for asking patients to fast before measurement of lipid profiles for cardiovascular risk prediction. Sources of Funding This work was funded by the Danish Medical Research Council and the Copenhagen County Foundation. These are public research funds with no right to approve or disapprove the submitted manuscript. Disclosures Dr Nordestgaard is a consultant for Abbott, BG Medicine, and AstraZeneca and has received lecture honoraria from AstraZeneca, Merck, Pfizer, Sanofi-Aventis, and Boehringer Ingelheim. The other authors report no conflicts.

10 256 Circulation November 11, 28 References 1. Rifai N, Warnick GR. Lipids, lipoproteins, apolipoproteins, and other cardiovascular risk factors. In: Burtis CA, Ashwood ER, Bruns DE, eds. Tietz Textbook of Clinical Chemistry and Molecular Diagnostics. 4th ed. Philadelphia, Pa: Elsevier Saunders; 26: Executive summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA. 21;285: De Backer G, Ambrosioni E, Borch-Johnsen K, Brotons C, Cifkova R, Dallongeville J, Ebrahim S, Faergeman O, Graham I, Mancia G, Manger C, V, Orth-Gomer K, Perk J, Pyorala K, Rodicio JL, Sans S, Sansoy V, Sechtem U, Silber S, Thomsen T, Wood D. European guidelines on cardiovascular disease prevention in clinical practice: Third Joint Task Force of European and Other Societies on Cardiovascular Disease Prevention in Clinical Practice. Eur Heart J. 23;24: Cohn JS, McNamara JR, Cohn SD, Ordovas JM, Schaefer EJ. Postprandial plasma lipoprotein changes in human subjects of different ages. J Lipid Res. 1988;29: Schaefer EJ, Audelin MC, McNamara JR, Shah PK, Tayler T, Daly JA, Augustin JL, Seman LJ, Rubenstein JJ. Comparison of fasting and postprandial plasma lipoproteins in subjects with and without coronary heart disease. Am J Cardiol. 21;88: Bansal S, Buring JE, Rifai N, Mora S, Sacks FM, Ridker PM. Fasting compared with nonfasting triglycerides and risk of cardiovascular events in women. JAMA. 27;298: Nordestgaard BG, Benn M, Schnohr P, Tybjaerg-Hansen A. Nonfasting triglycerides and risk of myocardial infarction, ischemic heart disease, and death in men and women. JAMA. 27;298: Agerholm-Larsen B, Tybjaerg-Hansen A, Frikke-Schmidt R, Gronholdt ML, Jensen G, Nordestgaard BG. ACE gene polymorphism as a risk factor for ischemic cerebrovascular disease. Ann Intern Med. 1997;127: Altman DG. Practical Statistics for Medical Research. 1st ed. London, UK: Chapman & Hall; 1991: Clarke R, Shipley M, Lewington S, Youngman L, Collins R, Marmot M, Peto R. Underestimation of risk associations due to regression dilution in long-term follow-up of prospective studies. Am J Epidemiol. 1999;15: Wilder LB, Bachorik PS, Finney CA, Moy TF, Becker DM. The effect of fasting status on the determination of low-density and high-density lipoprotein cholesterol. Am J Med. 1995;99: Zilversmit DB. Atherogenic nature of triglycerides, postprandial lipidemia, and triglyceride-rich remnant lipoproteins. Clin Chem. 1995; 41: Cohen JC, Noakes TD, Benade AJ. Serum triglyceride responses to fatty meals: effects of meal fat content. Am J Clin Nutr. 1988;47: Dubois C, Beaumier G, Juhel C, Armand M, Portugal H, Pauli AM, Borel P, Latge C, Lairon D. Effects of graded amounts ( 5 g) of dietary fat on postprandial lipemia and lipoproteins in normolipidemic adults. Am J Clin Nutr. 1998;67: Patsch JR, Miesenbock G, Hopferwieser T, Muhlberger V, Knapp E, Dunn JK, Gotto AM Jr, Patsch W. Relation of triglyceride metabolism and coronary artery disease: studies in the postprandial state. Arterioscler Thromb. 1992;12: Murphy MC, Isherwood SG, Sethi S, Gould BJ, Wright JW, Knapper JA, Williams CM. Postprandial lipid and hormone responses to meals of varying fat contents: modulatory role of lipoprotein lipase? Eur J Clin Nutr. 1995;49: Jeppesen J, Chen YD, Zhou MY, Wang T, Reaven GM. Effect of variations in oral fat and carbohydrate load on postprandial lipemia. Am J Clin Nutr. 1995;62: Pocock S, Shaper AG, Phillips AN. Concentrations of high density lipoprotein cholesterol, triglycerides and total cholesterol in ischaemic heart disease. BMJ. 1989;298: Talmud P, Hawe E, Miller G, Humphries S. Non-fasting apolipoprotein B and triglyceride levels as useful predictor of coronary heart disease risk in middle-aged UK men. Arterioscler Thromb Vasc Biol. 22;22: Thompson A, Danesh J. Associations between apolipoprotein B, apolipoprotein AI, the apolipoprotein B/AI ratio and coronary heart disease: a literature-based meta-analysis of prospective studies. J Intern Med. 26; 259: Sarwar N, Danesh J, Eiriksdottir G, Sigurdsson G, Wareham N, Bingham S, Boekholdt SM, Khaw KT, Gudnason V. Triglycerides and the risk of coronary heart disease: 1,158 incident cases among 262,525 participants in 29 Western prospective studies. Circulation. 27;115: Sharrett AR, Ballantyne CM, Coady SA, Heiss G, Sorlie PD, Catellier D, Patsch W, for the Atherosclerosis Risk in Communities Study Group. Coronary heart disease prediction from lipoprotein cholesterol levels, triglycerides, lipoprotein(a), apolipoproteins A-I and B, and HDL density subfractions: the Atherosclerosis Risk in Communities (ARIC) Study. Circulation. 21;14: Eberly LE, Stamler J, Neaton JD. Relation of triglyceride levels, fasting and nonfasting, to fatal and nonfatal coronary heart disease. Arch Intern Med. 23;163: Walldius G, Jungner I. Apolipoprotein B and apolipoprotein A-I: risk indicators of coronary heart disease and targets for lipid-modifying therapy. J Intern Med. 24;155: Zilversmit DB. Atherogenesis: a postprandial phenomenon. Circulation. 1979;6: Kolovou GD, Anagnostopoulou KK, Daskalopoulou SS, Mikhailidis DP, Cokkinos DV. Clinical relevance of postprandial lipaemia. Curr Med Chem. 25;12: CLINICAL PERSPECTIVE Lipid profiles are usually measured after fasting; however, it would be much simpler for patients if a random nonfasting sample could be used. We tested the hypotheses that lipid profiles change only minimally in response to normal food intake and that random nonfasting levels predict cardiovascular events. We cross-sectionally studied adults from the Copenhagen General Population Study and 9319 adults from the Copenhagen City Heart Study, of whom 1166 developed cardiovascular events during 14 years of follow-up. After normal food intake, individuals in the general population had a maximum mean change from fasting levels of.2 mmol/l ( 8 mg/dl) for total cholesterol at to 2 hours after the last meal,.2 mmol/l ( 8 mg/dl) for low-density lipoprotein cholesterol at to 2 hours,.1 mmol/l ( 4 mg/dl) for high-density lipoprotein cholesterol at to 5 hours, and.3 mmol/l (26 mg/dl) for triglycerides at 1 to 4 hours after the last meal. Highest versus lowest tertile of nonfasting total cholesterol, non high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, apolipoprotein B, triglycerides, ratio of total to high-density lipoprotein cholesterol, and ratio of apolipoprotein B to apolipoprotein A1 and lowest versus highest tertile of nonfasting high-density lipoprotein cholesterol and apolipoprotein A1 predicted 1.7- to 2.4-fold increased risk of cardiovascular events. Because we detected only minimal changes in levels of lipids, lipoproteins, and apolipoproteins in response to normal food intake in the general population, changes that are clinically unimportant, and because nonfasting levels predict cardiovascular events, our data challenge the necessity for asking patients to fast before measurement of lipid profiles for cardiovascular risk prediction.