Association of Postprandial Triglyceride and Retinyl Palmitate Responses With Asymptomatic Carotid Artery Atherosclerosis in Middle-aged Men and Women
The Atherosclerosis Risk in Communities (ARIC) Study
Abstract Blood lipid alterations after a fatty meal may be atherogenic, but there is little information regarding their associations with disease independent of fasting lipids. Asymptomatic atherosclerosis cases (n=229) and 373 control subjects free of atherosclerosis, as defined by carotid intima-media thickness on ultrasound images, were given a fatty meal with vitamin A, followed by 3.5- and 8-hour measurements of triglycerides (TGs), TG-rich lipoprotein TGs, apoproteinB48, and retinyl palmitate. Among white men and women but not among blacks, case status was associated with greater postprandial responses of TGs and TG-rich lipoprotein TGs, but only in nonobese persons (body mass index <30 kg/m2). The associations were strong and significant after controlling for coronary risk factors (odds ratio, ≈2.0) and fasting TGs (odds ratio, 1.5). Associations with other postprandial lipid measurements did not persist after controlling for fasting lipids. Elevated postprandial TGs appear to be an independent risk factor for carotid intimal thickening in nonobese whites. The lack of such a relation in obese subjects and the lipid profile they manifest suggest that postprandial TGs must be accompanied by accumulation of TG-rich lipoprotein remnants to be atherogenic.
Reprint requests to Dr A. Richey Sharrett, Epidemiology and Biometry Program, National Heart, Lung, and Blood Institute, National Institutes of Health, Rockledge Bldg Room 8164, Bethesda, MD 20892-7934. E-mail email@example.com.
- Received March 3, 1995.
- Accepted July 28, 1995.
Associations between atherosclerotic diseases and elevated fasting plasma LDL-C and reduced fasting plasma HDL-C are well established.1 However, many individuals without fasting lipid abnormalities develop atherosclerotic diseases, and several lines of evidence suggest that nonfasting lipid measurements may be more relevant to atherogenesis. Typical American diets are associated with measurable postprandial lipemia 18 hours per day.2 A number of studies have consistently shown that in the hours following a fatty test meal, plasma TGs are higher in men with CHD than in control subjects without CHD.3 4 5 6 7 8 9 10 CHD cases also usually have higher fasting TG levels than do control subjects, and given the high correlation between fasting and postprandial TG values, these studies have often been too small to evaluate the independent effects of postprandial TGs. Furthermore, many of these studies were not designed to select representative cases or comparable control subjects, and clinically recognized CHD, along with the medications and changes in diet and activity that accompany it, may affect TG responses. It is preferable, therefore, to evaluate associations between postprandial lipemia and asymptomatic atherosclerosis cases and comparable control subjects from the same population, as has been done in one small study of hypercholesterolemic subjects.11
The ARIC Group performed one of two collaborating case-control studies to evaluate the role of postprandial lipemia in atherosclerosis. From a population of individuals who were free of clinical atherosclerotic disease, ARIC investigators selected case and control subjects on the basis of ultrasound imaging evidence of extracranial carotid atherosclerosis. The other study, which was performed at Columbia University and reported in a companion article,12 selected clinical CHD cases prior to treatment and control subjects. Both studies used common protocols and measured the same lipids: plasma TGs, TGs in the d<1.006–g/mL fraction (containing TGRLs), and, as markers of intestinal lipoproteins, plasma RP and the ratio of apoB48 to apoB100 in the d<1.006–g/mL plasma fraction. The objectives of this two-study collaboration were to evaluate the associations between postprandial lipemia and atherosclerotic diseases after adjustment for the major coronary risk factors (including fasting plasma lipid and lipoprotein levels) and to compare two sets of these postprandial lipemia associations in those with carotid atherosclerosis and those with CHD.
The population for this study was drawn from the ARIC Study cohort, which was selected as a probability sample of 15 800 men and women between the ages of 45 and 64 years from four US communities. In three of these communities (Forsyth County, North Carolina; the suburbs of Minneapolis, Minn; and Washington County, Maryland), all age-eligible residents were included in the sampling frame. In the fourth community (Jackson, Miss), only black residents were sampled. Details of sampling and procedures are provided in manual 2 for visit 1 of the ARIC Study protocol.13
Participants in the postprandial study were selected from the ARIC cohort as cases of carotid atherosclerosis or as control subjects on the basis of measurements of carotid artery intima-media wall thickness on ultrasound images. These measurements provide an index of carotid atherosclerosis, which is strongly associated with fasting lipids, smoking, blood pressure, and other established CHD risk factors.14 15 Cases had either two measurements of maximum carotid artery wall thickness that were >2.5 mm or bilateral thickening, corresponding to a maximum thickness >1.7 mm at the internal carotid artery, 1.8 mm at the carotid bifurcation, or 1.6 mm at the common carotid segments. To supplement the number of case subjects who were black, criteria for this group were relaxed to 1.2, 1.675, and 1.0 mm, respectively, in the three carotid regions. These criteria resulted in case selection at approximately the 95th percentile wall thickness value for whites and at the 90th percentile for blacks. Control subjects in both ethnic groups had maximum wall thickness values <75th percentile in all carotid segments and in the popliteal artery. All cases and control subjects were required to meet criteria for minimum ultrasound visualization of artery wall boundaries.
After selection of these candidate cases and control subjects, individuals with the following self-reported manifestations of cardiovascular disease were excluded: a history of angina on effort; physician-diagnosed heart attack, transient ischemic attack, or stroke; or intermittent claudication. Cases were then matched pairwise with control subjects within strata defined by community, race, sex, and 10-year age group. For each case, a control subject with an ultrasound exam date close to that of the case subject was randomly chosen from among those with the highest level of wall boundary visualization.
The postprandial study was integrated into the second exam of the ARIC cohort at the four centers during 1990 through 1993. The study was approved by each institutional review board, and separate informed consent was obtained from each. Details of the study procedures have been described.16 Cases and control subjects were identified by the ARIC Coordinating Center, and lists of eligible participants were distributed as pairs of identification numbers. The identification of each participant as a case or control subject was not known to the field staff, laboratory staff, or participants themselves. Whenever feasible, both members of a case-control pair were scheduled for examination during the same week to allow simultaneous shipment of their blood samples to the lipid laboratory. A screening interview was conducted during recruitment to exclude pregnant women and individuals with specific medical conditions: hypertriglyceridemia (reported by either the participant or as a fasting TG level >4.52 mmol/L [400 mg/dL]); gall bladder disease; pancreatitis; intestinal malabsorption; chronic liver or kidney disease; use of insulin, hypoglycemic or lipid-lowering medications, thyroid preparations, β-blocking agents, or contraceptive hormones; and reported intolerance for dairy products or other constituents of the test meal.
Nonlipid variables such as anthropometric measures, fasting insulin level, medication use, blood pressure at rest, and smoking habits were collected by trained personnel according to the standardized ARIC Study protocol.13 Hypertension was defined as the mean of two blood pressures at rest that exceeded 140/90 or self-reported use of antihypertensive medication. A follow-up interview was conducted by telephone 3 days after the fat-tolerance test to document digestive problems suggestive of malabsorption.
On the day of the fat-tolerance test, participants arrived at the examination center between 7 and 8:30 am after fasting and abstaining from heavy physical work or exercise for 12 hours. A fasting blood specimen was drawn before the test meal was administered. The liquid test meal consisted of heavy whipping cream, ice cream, safflower oil, chocolate syrup, and powdered protein (Promod, Ross Laboratories) and contained 1265 kcal, 32 g protein, 48 g carbohydrate, 105 g fat (52 g saturated), 300 mg cholesterol, and 100 000 IU vitamin A (Aquasol, Armour Pharmaceutical Co) for each 2 m2 of body surface area, which was calculated as W0.425 × H0.725, where W is body weight in kilograms and H is body height in centimeters.17 Lactaid was added to the test meal as a precaution against lactose intolerance. Test meals were consumed within 15 minutes, after which time the participants were instructed to take nothing by mouth except water, unsweetened black coffee or tea, or sugarless soft drinks for 8 hours and to abstain from exercise and heavy physical work; they were, however, permitted to take prescribed medications. Blood specimens were drawn 3.5 and 8 hours after the participant began drinking the test meal. The limitation to only two postprandial blood collections has little affect on the classification of postprandial response. Correlations between responses that were calculated from two postprandial samples and from four postprandial samples (measured in our companion study12 ) exceeded .97 for all postprandial measures used. Participants remained at the examination center until the 3.5-hour-postprandial sample was drawn, then left, and returned for the 8-hour-postprandial collection. Medication use, consumption of any food or alcohol, or participation in heavy work or exercise during the test period were documented by interview before the 8-hour sample was taken.16
Postprandial blood specimens were collected into tubes containing EDTA (1.5 mg/mL blood). Plasma was separated by centrifugation (1500g, 20 minutes at 4°C), stored under N2 at 4°C for 1 to 3 days, and shipped on crushed ice to the central laboratory at the Baylor College of Medicine, Houston, Tex. Lipid measurements were performed within 7 days after arrival of the samples at the laboratory, with the exception of RP in plasma, which was stored under N2 in light-protected ampoules at −70°C and analyzed in batches.
Total plasma cholesterol and TGs were measured enzymatically on a Cobas-Fara centrifugal analyzer (Roche Diagnostics) with commercially available kits (catalog Nos. 236691 and 701912, respectively; Boehringer Mannheim Diagnostics). The HDL-C level was determined by measuring cholesterol content in the supernatant liquid after precipitation of the plasma with MgCl2 and dextran sulfate.18 LDL-C levels were calculated according to the formula of Friedewald et al.19 ApoB levels were determined by radioimmunoassay.20
TGRLs (d<1.006 g/mL) were isolated by ultracentrifugation from 5 mL of plasma of the fasting blood sample and of the 3.5- and 8-hour-postprandial specimens.21 Samples were spun for 18 hours at 10°C at 105 000g in a Beckman 50.3 rotor. TGs in the top fraction were obtained by tube slicing and were determined by the enzymatic procedure described above.
RP levels in plasma were determined by high-pressure liquid chromatography (high-pressure liquid chromatography pump model 501, tunable absorbance detector model 486, sample processor model 712, and data module model 743; all from Waters Instruments) as previously described.22
The apoB48-apoB100 ratio in TGRLs was determined exactly as described previously.22 In brief, aliquots of the d<1.006–g/mL fraction were subjected to SDS gel electrophoresis on a 3% stacking gel and 5% separating gel. The gels were stained with 0.25% (vol/vol) Coomassie Brilliant Blue G-250 for 1 hour and destained in 40% (vol/vol) methanol and 10% (vol/vol) acetic acid, and relative abundances of apoB48 and apoB100 were determined by scanning the gels with a laser densitometer for 20 hours. The apoB48-apoB100 ratio rather than the absolute level of apoB48 is reported, because the ratio is measured more reliably with this method.
ApoE genotypes were determined by restriction isotyping. Genomic DNA was extracted from buffy coats, amplified by the polymerase chain reaction23 using the primers reported by Emi et al,24 and restricted with Hha I, as suggested by Hixson and Vernier.25 Digested DNA was then subjected to electrophoresis on 12% polyacrylamide gels under nondenaturing conditions. ApoE polymorphisms were typed from the ethidium bromide–stained gels.
LDL particle size was determined by gradient gel electrophoresis using 2% to 16% nondenaturing polyacrylamide-agarose gels (PAA 2% to 16%, Pharmacia). Aliquots of plasma that had been stored for no longer than 1 week at 4°C were mixed with a solution of 40% (vol/vol) sucrose and 0.1% (vol/vol) bromophenol blue and subjected to electrophoresis in a Pharmacia GE 2/4 apparatus at 120 V for 18 hours at 10°C using 0.09 mol/L Tris, 0.08 mol/L boric acid, and 0.003 mol/L Na2EDTA, pH 8.3, as the running buffer. Gels were stained with Sudan black B for 20 hours, destained in a 50% (vol/vol) Cellosolve solution (Sigma) for 2 to 3 days, and stored in 25% (vol/vol) Cellosolve26 to restore gel size and shape before scanning on an Ultrascan XL laser densitometer equipped with the gsxl software package (LKB Instruments Inc). Two plasma standards, each derived from a single donor and containing either small, dense or large, buoyant LDL, as determined by zonal ultracentrifugation,27 were analyzed with all unknown plasma specimens. LDL particle size was assigned to subclass pattern A, B, or I according to the nomenclature of Krauss and Burke.28 Assignment was made by three independent observers. For quality-control purposes, approximately 5% of samples were reanalyzed. All samples subjected to repeated analyses showed complete agreement with their original assignment.
For internal quality control of lipid and apoprotein measurements, control plasma pools were used. The CVs were 2.7% for TGs, 3.7% for HDL-C, 5.2% for LDL-C, 13% for RP, and 9% for apoB. External quality control consisted of our laboratory’s participation in the Centers for Disease Control and Prevention Lipid Standardization Program. In addition, aliquots from a subset of samples were taken from the blood collection tubes, stored at each field center for an additional week, and sent to the central laboratory in the subsequent weekly mailing in a blinded fashion, thereby providing a measure of overall variability (ie, variability due to storage, shipping, sample processing, transcription, and analyses). The CVs for these blinded replicates for TGs, RP, LDL-C, and HDL-C were 15%, 18%, 8%, and 11% respectively, with corresponding correlations of .95, .93, .85, and .95.
The fat content of the test meal was measured after extraction according to a modification29 of the method of Folch et al30 and gravimetric determination after evaporation of the organic solvents. Fat content averaged 0.266 g/mL and varied little among 153 test meals analyzed (SD, 0.018 g/mL; CV, 6.8%).
Comparability between the Columbia and Baylor laboratories was assessed by blinded parallel analyses of aliquots from the same postprandial samples. The correlation between values obtained at the two laboratories was .90 for RP (54 pairs of measurements) and .99 for TGs (37 pairs). Compared with Baylor measurements, Columbia values were 10% higher for RP and 4% higher for TGs.
Means and proportions for cases can be interpreted as estimates for the entire population of cases who met the selection criteria for this study (as we included a random sample of such cases and with the assumption that refusals were also random). Corresponding control values can be interpreted as adjusted to the case distribution of community, race, sex, and 10-year age group. Likewise, case-control comparisons can be interpreted as adjusted for these variables.
The primary analysis of the association between postprandial lipids and case-control status used race-specific conditional logistic regression models, which were based on strata defined by community, sex, age group, and 6-month period of examination. Formally, these are models of probability of “caseness” (carotid atherosclerosis) as a function of the level of postprandial lipids, controlling by design for the variables defining the conditioning strata and other covariates in the model.31
The postprandial lipid values in these models are AUCs of lipid measurement above baseline plotted versus time, as approximated by the three time points available. AUCs for TGs can be interpreted as time (8 hours) multiplied by the difference between the fasting TG value and its mean over the 8-hour period, with negative values set to zero. The AUC was found to discriminate between cases and control subjects better than other indicators of response, such as 8-hour values. Modifications of associations according to sex, smoking status, hypertension, LDL-C, BMI, and other factors that are known to be related to postprandial lipemia or to carotid atherosclerosis were evaluated by inclusion of appropriate interaction terms in the models.
A total of 528 white and 124 black ARIC participants met carotid wall thickness criteria as cases (Table 1⇓), and 326 whites and 75 blacks were invited to participate in this study as cases or control subjects. An additional 456 blacks met the relaxed case criteria and 110 of them were invited to participate. The pool of eligible control subjects was much larger. Of the potential cases, 18% to 25% were excluded on the basis of evidence of cardiovascular disease, as were 7% to 11% of potential control subjects. Refusal rates were ≤8%. Other conditions that resulted in exclusion were found in 37% to 49% of potential cases and 19% to 26% of potential control subjects. A few study participants had test-meal intolerance or failed to consume the full test meal. Remaining for study were 170 white and 59 black cases and 274 white and 99 black control subjects.
Table 2⇓ shows the sex, age, and mean carotid intima-media thickness of white and black cases and control subjects and their risk factor levels after adjustment for age, sex, community, and date of examination. As designed, within each racial group cases and control subjects were similar with respect to sex and age. In whites, mean carotid wall thickness was 1.19 mm in cases and 0.63 mm in control subjects, a difference of 0.56 mm. The maximum value for mean carotid wall thickness for cases was 2.26 mm; only 9% had a mean thickness exceeding 1.5 mm, and 13% had carotid stenosis, defined as any lumen narrower than 2 mm.
For blacks, the difference between cases and control subjects was only 0.35 mm because of the relaxed case criteria used. Only 25 black participants qualified as cases on the basis of case criteria used for whites, too few for separate analysis.
Compared with control subjects, white cases had higher mean fasting LDL-C and TG levels and lower HDL-C levels. More were current smokers, more had hypertension, and mean BMI was greater in cases. Among blacks there were no significant case-control risk factor differences; although differences were generally in the expected direction, but fasting TG levels were somewhat lower in black cases than control subjects.
Analyses of postprandial lipids were performed separately for each race. Mean fasting and postprandial levels of TGs, TGRL TGs, RP, and percent of apoB48, adjusted for age, sex, community, and date of initial examination, are shown for white participants in Figs 1 through 4⇓⇓⇓⇓. In cases, TG levels rose from a fasting mean of 1.66 mmol/L to 3.45 mmol/L at 3.5 hours after the test meal and declined to 2.57 mmol/L at 8 hours (Fig 1⇓). TGRL TGs rose from 1.06 mmol/L in fasting plasma to 2.53 mmol/L at 3.5 hours and 1.74 mmol/L at 8 hours (Fig 2⇓). For both total TGs and TGRL TGs and for each of the three time periods, the levels were substantially and significantly (P<.0001) higher in white cases than control subjects. RP levels, usually less than the detection limit of 0.5 μg/dL in fasting plasma, rose in white cases to 84 μg/dL at 3.5 hours and 127 μg/dL at 8 hours (Fig 3⇓). Levels in control subjects were similar to those in cases in fasting and 3.5-hour plasma samples but were lower at 8 hours (P<.01). The apoB48 concentration in fasting plasma was approximately 3% of the TGRL apoB100 concentration in white cases and control subjects, increased to 6% in the 3.5-hour plasma, and declined in the 8-hour plasma to 4.4% in cases and to 3.9% in control subjects (P<.05 for the difference at 8 hours; Fig 4⇓). Mean fasting and postprandial levels of TGs, TGRL TGs, RP, and apoB48 percent were also examined for blacks. No significant case-control differences were found.
Independent associations of postprandial AUCs with case-control status were evaluated in analyses that controlled for fasting TGs; risk factors; and the selection factors of age, sex, community, and date of examination. Potential interactions involving sex, smoking status, hypertension, diabetes, serum insulin level, LDL-C (>4.13 mmol/L [160 mg/dL]), waist-hip ratio (greater than the median), and BMI (>30 kg/m2) were tested. Only the AUC interaction with BMI was consistent in both races and significant or nearly significant (P<.10) in these models. Average AUCs with their SDs in the population were 8.04±5.31 (mmol/L)×hours for TGs, 6.10±4.85 (mmol/L)×hours for TGRL TGs, and 580±306 (μg/dL)×hours for RP. The ORs in Table 3⇓ correspond to approximate 1-SD–higher values of the AUCs, ie, 500 and 450 (mg/dL)×hours (5.64 and 5.08 [mmol/L]×hours, respectively) for TGs and TGRL TGs and 300 (μg/dL)×hours for RP. For white participants, these ORs were 1.22 for TGs, 1.28 for TGRL TGs, and 1.03 for RP (none significant). However, ORs differed by BMI group for TGs and TGRL TGs (each interaction significant at P<.05) and for RP (P=.09 for interaction) and were consistently higher in the BMI<30 kg/m2 group. In this BMI group, comprising 81% of white participants, ORs were substantial and statistically significant for TGs (1.47) and for TGRL TGs (1.52) but not for RP (1.12). Corresponding ORs were <1.0 (not significant) for the BMI≥30 kg/m2 group.
A similar pattern of AUC-BMI associations was observed in black subjects (Table 3⇑), but none of the ORs were significant. ORs for the BMI<30 kg/m2 group were 1.22 for TGs and 1.25 for TGRL TGs, with corresponding ORs <1.0 for the BMI≥30 kg/m2 group. No substantial associations were seen for RP.
Control for fasting HDL, in addition to fasting TGs and the covariates listed in Table 3⇑, had little effect on the significant associations reported. ORs for TGs in whites with a BMI <30 kg/m2 were reduced from 1.47 to 1.44 after inclusion of HDL-C. For TGRL TGs the ORs were reduced from 1.52 to 1.49.
There were no significant interactions between apoE genotype and TG AUC effects, but ORs tended to be higher for those with the 4/3 genotype (OR, 2.37 in whites and 1.60 in blacks; Table 4⇓). Only the OR for apoE genotype 4/3 in whites was statistically significant.
Fasting TGs were consistently higher in the BMI≥30 kg/m2 group than in the BMI<30 kg/m2 group, by 41% in white control subjects (P<.001), 16% in white cases (P<.05), 10% in black control subjects (NS), and 3% in black cases (NS). Also, TGRLs appeared to be larger in the BMI≥30 kg/m2 group. A linear model regressing fasting TGRL TGs against BMI group, TGRL apoB, age, race, and sex showed a 0.28–mmol/L higher TGRL TG for any fixed apoB level in the BMI≥30 kg/m2 group than in thinner subjects (P<.0001). In contrast, a similar analysis of denser lipoproteins showed only a 0.02–mmol/L higher TG value in the BMI≥30 kg/m2 group (NS). Thus, TGRLs in the BMI≥30 kg/m2 group were more TG enriched than in the BMI<30 kg/m2 group.
Because BMI modified the association between postprandial TGs and case-control status, it was important to study whether BMI also modified the associations between postprandial TGs and either HDL-C level or LDL size. The results indicated that BMI did not significantly modify these associations. This was examined by using two linear regression models. The first related TG AUC to age, sex, BMI group, HDL, and the interaction term (BMI group×HDL). This model, which was run separately in white cases and control subjects, showed that higher postprandial TGs were consistently related to lower fasting HDL-C values, but these associations were not significantly or consistently modified by BMI in the two groups. The second model related TG AUC to age, sex, case-control status, BMI group, LDL size, and the interaction term (BMI group×LDL size) and was run in white cases and control subjects together. Postprandial TGs were associated with smaller LDL particle sizes, but BMI interactions with this association were not significant.
This study shows an association between postprandial TGs and carotid atherosclerosis, as defined by measures of intima-media thickness, that is independent of fasting lipids and other risk factors in nonobese, middle-aged white subjects. Unfortunately, there were too few cases with sufficient carotid thickening for an adequate test of this hypothesis in blacks. This is the first report from a study with several features that were designed to assure case-control comparability: selection of cases and control subjects from the same community-based population samples; matching on key variables; blinded simultaneous assays of case and control plasma; and lack of participant awareness of their case-control status, making them unlikely to differ in any dietary or other health behaviors related to plasma lipid levels. Care was taken to reduce bias by excluding persons who were using any medication or having any medical condition that might have selectively affected postprandial lipemia. Because cases were asymptomatic, with mean intima-media thicknesses ≤2 mm, the findings of this study pertain to the early stages of carotid atherosclerosis. Thus, this study confirms, but only in nonobese whites, previous studies that have shown associations between elevated postprandial TG levels and cardiovascular disease and provides evidence that postprandial TG elevation is not a consequence of the disease, its medications, or its lifestyle changes but is more likely involved in atherogenesis.
A 1-SD–greater postprandial AUC for TGs or TGRL TGs conveyed ORs of approximately 1.5 in nonobese whites, after adjustment for LDL-C and fasting TGs (Table 3⇑), clear evidence of independent associations with carotid disease. Associations that were adjusted for fasting TGs, however, may underestimate the pathogenic importance of postprandial lipids: remnants of intestinal and hepatic TGRLs occupy very similar metabolic pathways, for example, and postprandial lipid metabolism is complex. Associations of postprandial TGs with disease can also be studied by using the 8-hour post–test meal TG level (rather than the post–test meal TG increase) as the independent variable and not including fasting TG as a covariable. When this is done, associations with disease are strengthened, with ORs for 1-SD elevations in nonobese whites (in the same model used in Table 3⇑) of 2.01 for TGs (P<.0001) and 2.04 for TGRL TGs (P<.0001). Thus, post–test meal TGs are independently associated with disease in this group, as shown in Table 3⇑, and the net association may be strong, as discussed in this paragraph. More informative estimates of the strength of that association must await more specific measures of atherogenic lipid changes that occur postprandially.
Finding substantially lower ORs in the BMI≥30 kg/m2 group than in thinner participants, despite elevated fasting TG levels in the former, was unexpected. These subjects were given more of the test meal, as the quantity was scaled to body surface area. If the quantity were inappropriately large relative to their usual dietary intake, this could have led to an atypical postprandial response unrelated to the extent of their carotid atherosclerosis. We do not believe this to be the explanation for the lower OR in the BMI≥30 kg/m2 group. Body surface area is believed to be approximately proportional to a person’s caloric need,32 and scaling by body surface area is often used in postprandial lipid research.8 10 11 33 34 As a result, an average test meal quantity was only 13% greater in the BMI≥30 kg/m2 subjects. Furthermore, Patsch et al10 also found postprandial lipemia to be strongly associated with CHD, despite their use of a test meal that contained 30% more fat than in the present study.
Our companion study12 confirmed a lower OR in persons with a greater BMI. This replication of results justifies speculation regarding possible BMI effects. Elevated fasting TGs in persons with a greater BMI concur with studies showing that even normolipidemic obesity is associated with increased VLDL TG production rates.35 36 37 Increased VLDL TG production is expected to augment postprandial lipemia, because intestinal and hepatic TGRLs compete for a common clearance pathway. Lewis et al38 found the increment in postprandial TGs to be threefold greater in obese than control subjects, while the retinyl ester increment was only 1.6-fold greater, suggesting a greater role for hepatic TG secretion in postprandial triglyceridemia in obesity.
How then can the impact of BMI on the association of postprandial TGs with atherosclerosis be explained? Current concepts suggest that postprandial triglyceridemia per se is not harmful but serves as a marker of complex processes that may ultimately promote atherosclerosis. When the integrated level of TGRLs in the postprandial state is high, CETP-mediated transfer of CE from HDL and LDL to TGRLs in exchange for TGs is extensive. Hence, HDL-C levels decrease, LDL particle size is reduced, and CE-rich remnants of TGRLs are formed. Each of these consequences of postprandial TG metabolism is potentially atherogenic.
It is conceivable that the postprandial response in obese persons may be atypical with respect to any of the consequences described above. Our data show no influence of BMI on the relation between postprandial TG and HDL-C levels, suggesting that CETP-mediated transfer of lipids between HDL and TGRLs does not differ. Similarly, BMI appears to have no significant effect on the relation between postprandial TGs and LDL particle size. Ginsberg et al39 also report that BMI does not modify the associations between postprandial TGs and either HDL-C level or LDL particle size. However, fasting TGRLs had a higher TG content for any fixed apoB level in the BMI≥30 kg/m2 group, indicating that for equal TG levels, this group had fewer TGRL particles than the BMI<30 kg/m2 group.
Thus, in obese individuals, hepatic secretion of larger, less dense TGRLs may abolish the marker function of postprandial TG measurements. If this conclusion were confirmed in other studies, our findings would imply that an impaired TG metabolism is atherogenic because of its association with TGRL remnant accumulation rather than its effects on LDL particle size or HDL-C levels. Differences in TG metabolism with levels of obesity may also explain the general inconsistency in results of studies on TGs as a risk factor for cardiovascular diseases.40
Because our companion study showed clear differences in the postprandial TG association with CHD by sex, we examined the same associations in relation to carotid atherosclerosis. No sex interactions were significant. Among whites, we found a pattern similar to that in the Columbia study, ie, somewhat increased ORs in men. For TGRL TGs, ORs were 1.36 in men and 1.09 in women, and in the BMI<30 kg/m2 group, 1.56 and 1.34, respectively. The pattern was reversed in blacks, however: 0.84 in men and 1.31 in women.
Postprandial lipoprotein associations may be stronger for CHD than for early atherosclerosis. Whites with a BMI<30 kg/m2 in our study had an OR of 1.47 for carotid atherosclerosis for each 500 (mg/dL)×hours (5.64 [mmol/L]×hours) increase in TG AUC (Table 3⇑), compared with an OR of 2.3 for CHD in men in the Columbia study that used a similar analysis. We recently reported41 a higher fasting LDL-C–apoB ratio in ARIC subjects with asymptomatic carotid thickening than in comparable subjects with clinical CHD. We suggested that the difference might be due to a greater role for elevated TGs in the CHD cases, as TGs may be thrombogenic and elevated TG levels are associated with lower LDL-C–apoB ratios. If the greater role for postprandial lipemia in CHD than in early atherosclerosis were confirmed, it might suggest, as we earlier hypothesized, that disorders of TG metabolism may have a greater influence on the later atherothrombotic processes associated with clinically apparent ischemic coronary disease than on early arterial thickening, as studied here.
Selected Abbreviations and Acronyms
|ARIC||=||Atherosclerosis Risk in Communities|
|AUC(s)||=||area(s) under the curve|
|BMI||=||body mass index|
|CETP||=||cholesteryl ester transfer protein|
|CHD||=||coronary heart disease|
|CV(s)||=||coefficient(s) of variation|
This research was carried out as a collaborative study supported by grant No. UO1 HL45467 and contracts N01-HC55015, N01-HC55016, N01-HC55018, N01-HC55019, N01-HC55020, N01-HC55021, and N01-HC55022 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md. We are indebted to the following for their assistance: Bowman Gray School of Medicine, Winston-Salem, NC: Regina deLacy, Delilah Cook, Carolyn Bell, Theresa Crotts, and Suzanne Pillsbury; The Johns Hopkins University, Baltimore, Md: Joel G. Hill, Patricia M. Crowley, Joyce B. Chabot, and Patricia Hawbaker; Methodist Hospital, Houston, Tex: Val Creswell, Julita Samoro, Wanda Wright, and Karima Ghazzaly; University of Minnesota, Minneapolis: Barbara Kuehl, Anne Murrill, Bryna Lester, and Jennifer Hill; University of Mississippi Medical Center, Jackson: Cora Walls, Dorothy Washington, Mattye Watson, and Nancy Wilson; University of North Carolina, Chapel Hill: Carol Summers, Catherine Burke, Deanna Horwitz, Carmen Woody, Peter DeSaix, Mal Foley, Tom Goodwin, and Richard Hayes.
Wallace RB, Anderson RA. Blood lipids, lipid-related measures, and the risk of atherosclerotic cardiovascular disease. Epidemiol Rev. 1987;9:95-119.
Barritt DW. Alimentary lipaemia in men with coronary artery disease and in controls. Br Med J. 1956;2:640-644.
Groot PHE, van Stiphout WAHJ, Krauss XH, Jansen H, van Tol A, van Ramshorst E, Chin-On S, Hofman A, Cresswell SR, Havekes L. Postprandial lipoprotein metabolism in normolipidemic men with and without coronary artery disease. Arterioscler Thromb. 1991;11:653-662.
Patsch JR, Miesenbock G, Hopferwieser T, Muhlberger V, Knapp E, Dunn JK, Gotto AM, Patsch W. Relation of triglyceride metabolism and coronary artery disease: studies in the postprandial state. Arterioscler Thromb. 1992;12:1336-1345.
Ryu JE, Howard G, Craven TE, Bond MG, Hagaman AP, Crouse JR. Postprandial triglyceridemia and carotid atherosclerosis in middle-aged subjects. Stroke. 1992;23:823-828.
Ginsberg HN, Jones J, Blaner WS, Thomas A, Karmally W, Fields L, Blood D, Begg M. Postprandial triglyceride and retinyl ester response to a fat-meal are independent predictors of coronary artery disease in middle-aged men, but not women. Arterioscler Thromb Vasc Biol. 1995;15:1829-1838.
National Heart, Lung, and Blood Institute. ARIC Study Operations Manual No 2: Cohort Component Procedures, Version 2.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1988.
Heiss G, Sharrett AR, Barnes R, Chambless LE, Szklo M, Alzola C and the ARIC Investigators. Carotid atherosclerosis measured by B-mode ultrasound in populations: associations with cardiovascular risk factors in the ARIC study. Am J Epidemiol. 1991;134:250-256.
Folsom AR, Eckfeldt JH, Weitzman S, Ma J, Chambless LE, Barnes RW, Cram KB, Hutchinson RG. Relation of carotid artery wall thickness to diabetes mellitus, fasting glucose and insulin, body size, and physical activity. Stroke. 1994;25:66-73.
National Heart, Lung, and Blood Institute. ARIC Study Postprandial Lipoproteins and Atherosclerosis: Operations Manual, Version 2.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1991.
Lohman TG, Roche AF, Martorell R. Anthropometric Standardization Reference Manual. Champaign, Ill: Human Kinetic; 1988.
Warnick GR, Benderson JM, Albers JJ. Dextran sulfate-Mg2+ precipitation procedure for quantification of high-density-lipoprotein cholesterol. Clin Chem. 1982;28:1379-1388.
Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein in plasma without use of the preparative ultracentrifuge. Clin Chem. 1972;18:499-502.
Schonfeld G, Lees RS, George PK, Pfleger B. Assay of total plasma apolipoprotein B concentration in human subjects. J Clin Invest. 1974;53:1458-1467.
Lipid Research Clinics Program. Manual of Operations: Lipid and Lipoprotein Analyses. Bethesda, Md: US Dept of Health and Human Services, Public Health Service, National Institutes of Health; 1982.
Saiki RK, Gelfand DH, Stoffel S, Scharf S, Higuchi R, Horn GT, Mullis KB, Erlich HA. Primer-directed amplification of DNA with thermostable DNA polymerase. Science. 1988;239:487-491.
Hixson JE, Vernier DT. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J Lipid Res. 1990;31:545-548.
McNamara JR, Campos H, Ordovas JM, Peterson J, Wilson PWF, Schaefer EJ. Effect of gender, age, and lipid status on low density lipoprotein subfraction distribution: results from the Framingham Offspring Study. Arteriosclerosis. 1987;7:483-490.
Krauss RM, Burke DJ. Identification of multiple subclasses of plasma low density lipoproteins in normal humans. J Lipid Res. 1982;23:97-104.
Carlson LA. Determination of serum triglycerides. J Atheroscler Res. 1963;3:334-336.
Folch J, Lees M, Sloane-Stanley GH. A simple method for the isolation and purification of total lipids from animal tissues. J Biol Chem. 1957;226:497-509.
Prentice RL, Pike R. Logistic disease incidence models and case-control studies. Biometrika. 1979;66:403-411.
Kleiber M. The Fire of Life: An Introduction to Animal Energetics. Huntington, NY: Robert E Krieger Publishing Co; 1975.
Genest J, Sniderman A, Cianflone K, Teng B, Wacholder S, Marcel Y, Kwiterovich P. Hyperapobetalipoproteinemia: plasma lipoprotein responses to oral fat load. Arteriosclerosis. 1986;6:297-304.
Weintraub MS, Rosen Y, Otto R, Eisenberg S, Breslow JL. Physical exercise conditioning in the absence of weight loss reduces fasting and postprandial triglyceride-rich lipoprotein levels. Circulation. 1989;79:1007-1014.
Grundy SM, Mok HYI, Zech L, Steinberg D, Berman M. Transport of very low density lipoprotein triglycerides in varying degrees of obesity and hypertriglyceridemia. J Clin Invest. 1979;63:1274-1283.
Ginsberg HN, Jones J, Begg M. Postprandial lipemia predicts the presence of coronary artery disease in normal weight but not obese men. Circulation. In press.
Sharrett AR, Patsch W, Sorlie PD, Heiss G, Bond MG, Davis CE. Associations of lipoprotein cholesterols, apolipoproteins A-I and B, and triglycerides with carotid atherosclerosis and coronary heart disease: the Atherosclerosis Risk in Communities (ARIC) Study. Arterioscler Thromb. 1994;14:1098-1104.