Genetic Effects on Postprandial Variations of Inflammatory Markers in Healthy Individuals
Circulating levels of inflammatory markers predict the risk of cardiovascular disease (CVD), mediated perhaps in part by dietary fat intake, through mechanisms only partially understood. To evaluate post-fat load changes in inflammatory markers and genetic influences on these changes, we administered a standardized high-fat meal to 838 related Amish subjects as part of the Heredity and Phenotype Intervention (HAPI) Heart Study and measured a panel of inflammatory markers, including C-reactive protein (CRP), interleukin-1β (IL-1β), matrix metalloproteinase-1 and -9 (MMP-1 and MMP-9), and white blood cell (WBC) count, before and 4 h after fat challenge (CRP prechallenge only). Heritabilities (h2 ± s.d.) of basal inflammatory levels ranged from 16 ± 8% for MMP-9 (P = 0.02) to 90 ± 7% for MMP-1 (P < 0.0001). Post-fat load, circulating levels of WBC, MMP-1, and MMP-9 increased by 16, 32, and 43% (all P < 0.0001), with no significant changes in IL-1β. Postprandial changes over the 4-h period were modestly heritable for WBC (age- and sex-adjusted h2 = 14 ± 9%, P = 0.04), but the larger MMP-1 and MMP-9 changes appeared to be independent of additive genetic effects. These results reveal that a high-fat meal induces a considerable inflammatory response. Genetic factors appear to play a significant role influencing basal inflammatory levels but to have minimal influence on post-fat intake inflammatory changes.
Levels of inflammatory markers in circulation predict cardiovascular disease (CVD) events (1,2), consistent with the emerging views of a strong inflammatory component to the pathogenesis of atherosclerosis and the transition to its clinical manifestations (3). Immune cells and their effector molecules are implicated in a number of pathogenic mechanisms in atherosclerosis, including endothelial dysfunction, leukocyte migration, extracellular matrix degradation, and platelet activation (3,4,5,6). For example, recent evidence suggests that proinflammatory cytokines, such as interleukin (IL)-6 and −1β, can mediate the migration of leukocytes to vascular endothelium and induce the expression of acute-phase proteins (7). The expression and production of matrix metalloproteinase-1 and -9 (MMP-1 and MMP-9) by macrophages in the plaques have been suggested to play a critical role in plaque rupture via their ability to degrade extracellular matrix (4).
Circulating levels of several inflammatory cytokines are known to increase following dietary fat intake (8,9,10,11,12,13,14,15), although the potential contribution of the postprandial inflammatory response to CVD risk is not well established. The mechanisms underlying the postprandial inflammatory response are unclear, although some evidence suggests that postprandial hypertriglyceridemia from a high-fat meal activates endothelial cells and initiates inflammatory cytokine production (16,17). Dietary fat intake induced expression of several inflammatory genes in mice (18,19). The degree to which genetic factors contribute to postprandial inflammation is unknown, although a genetic contribution to nonpostprandial circulating levels of inflammatory markers is well documented, with estimated heritabilities of 20–40% for white blood cell (WBC) and C-reactive protein (CRP) concentrations (20,21,22).
Given emerging evidence that the postprandial response in circulating lipids and/or inflammation factors may be associated with CVD risk (23,24), we have measured postprandial changes of inflammatory markers in response to a standardized high-fat challenge in subjects from large Amish families for the purpose of: (i) identifying factors associated with baseline levels and postprandial changes in inflammatory markers; and (ii) assessing the genetic contributions to the interindividual variation in baseline levels and postprandial changes in inflammatory markers. Inflammatory markers measured in this study include WBC, CRP, IL-1β, MMP-1, and MMP-9, all of which have been linked to various pathogenic stages of atherosclerosis or CVD risk in animal and clinical studies (1,2,4,7). Understanding the relative genetic and environmental contributions to inflammation may lead to the identification of modifiable factors that could shape individual risk and provide a future direction for CVD prevention.
Methods and Procedures
The Amish Heredity and Phenotype Intervention (HAPI) Heart Study began in 2002 to identify novel loci that interact with specific environmental exposures to modify risk factors for CVD. Subjects recruited for this study were ≥20 years old and were excluded if they had severe hypertension (blood pressure >180/105 mm Hg), malignancy, and kidney, liver, or untreated thyroid disease (25). This report includes the 838 subjects (456 men and 382 women) who completed the high-fat feeding intervention arm of the HAPI Heart Study. Subjects were also excluded from this arm if they had malabsorption disorders, lactose intolerance, gall bladder disease, or history of pancreatitis. Virtually all subjects can be connected into a single 14 generation pedigree (26). Our sample included 326 parent-offspring pairs, 633 sib pairs, 11 grandparent-grandchild pairs, 443 avuncular pairs, and 191 first cousin pairs that were informative for estimating heritabilities.
All subjects discontinued vitamins, supplements, and prescription medications beginning 7 days before participation. Subjects underwent a physical examination, including height and weight, using standard methods by a registered nurse during their first clinic visit at the Amish Research Clinic (Strasburg, PA). Blood pressure was measured in triplicate in the sitting position after the subject had been sitting quietly for 5 min by use of a standard sphygmomanometer, and the average of the measurements was calculated. Smoking status and medical history (including diabetes, heart attack, stroke, and cancer status) were assessed by questionnaires administered by trained study nurses. Fasting lipid profile (including total cholesterol, high-density lipoprotein cholesterol, and triglyceride) was assayed by the Quest Diagnostics (Horsham, PA). The protocol was approved by the Institutional Review Board at University of Maryland. Informed consent, including permission to contact relatives, was obtained before participation.
A high-fat meal was administered in the morning following an overnight 10-h fast at the Amish Research Clinic. This meal was a milkshake comprising heavy whipping cream, skim milk powder, and corn syrup, standardized to include 782 calories/m2 of the subject's body surface area. The total energy content per 100 g of milkshake was 26.6 g total fat (77.6% of total calories), 97.8 mg cholesterol, 2.4 g protein (3.1% of total calories), and 14.8 g carbohydrates (19.2% of total calories). An indwelling angiocatheter was placed in the right antecubital vein for serial blood draws before milkshake consumption (time 0) and at 1, 2, 3, 4, and 6 h after ingestion. Blood was processed within 1–2 h and serum frozen at −80 °C until the lab assay.
Assessment of inflammatory biomarkers
Blood samples collected at 0 (T0) and 4 h (T4) after the high-fat challenge were assayed for inflammatory biomarkers. The number of samples measured at baseline ranged from 779 for IL-1B to 837 for WBC. Due to budgetary constraints, only a subset of samples was measured at the 4-h time point, ranging from 61% (476/779) for IL-1B to 98% (822/837) for WBC. Serum concentrations of IL-1β were measured in triplicate and MMP-1 and MMP-9 concentrations in duplicate using an enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN) by the University of Maryland Cytokine Biochemistry Core Laboratory (Baltimore, MD). The means of the replicate values were used for data analyses. The assay detection ranges were 0.781–50 pg/ml for IL-1β, 0.156–10 ng/ml for MMP-1, and 31.2–2,000 ng/ml for MMP-9. Values above and below the range were assigned the maximum and minimum values of detection, respectively. The intra-assay coefficients of variation were 5.5, 7.5, and 5.8%, for IL-1β, MMP-1, and MMP-9, respectively. WBC counts were determined by complete blood count using Coulter LH750 Hematology Analyzer (Beckman Coulter, Fullerton, CA), and high-sensitive CRP (interassay coefficients of variation = 4.8%) were determined by endpoint nephelometry, as assayed by Quest Diagnostics.
All baseline inflammatory marker values were natural logarithm transformed to remove skewness. The differences between the (untransformed) T4 and T0 values (i.e., T4-T0) were approximately normally distributed and were tested for deviation from 0 by a paired t-test. Partial correlations (r) between inflammation markers and covariates were computed by first estimating the proportionate reduction in the variance of the trait (e.g., inflammatory marker) in the model associated with inclusion of the covariate of interest and then taking the square root of this quantity, with the direction of correlation assigned based on the sign of the covariate beta value. In order to adjust for potential nonlinear effect of age and allow the effect of age to vary by sex, correlation estimates were adjusted for age, age2, sex, and the corresponding age-by-sex interactions. Trait heritability (h2) was defined as the proportion of the total trait variance attributable to additive genetic effects and was estimated by modeling the phenotypic covariance (conditional upon covariate effects) between any two individuals in the pedigree as a function of their degree of biological relatedness. Statistical analyses were conducted using variance component analyses as implemented in Sequential Oligogenic Linkage Analysis Routines to account for familial relatedness of the data (version 4.0.7; Southwest Foundation for Biomedical Research, San Antonio, TX) (27).
Our sample size provided at least 42, 76, 95, and 99% power to detect heritabilities of 10, 20, 30, and 40%, respectively, for basal inflammatory markers (at α = 0.05). For postprandial inflammatory changes (including WBC, MMP-1, and MMP-9), our sample provided at least 36, 63, 87, and 97% power to detect heritabilities of 10, 20, 30, and 40%, respectively.
Baseline anthropometric and clinical characteristics
Baseline characteristics are presented in Table 1. The average age of study subjects was 44 (±14) years, and 54.4% of the participants were male. Very few among this study were previously diagnosed with myocardial infarction, stroke, or diabetes. Whereas none of the female participants were current smokers, 20% of the men reported current use of a pipe, cigar, and/or cigarettes. Overall, the study population represented a group of relatively healthy Amish adults.
Table 1. Baseline characteristics of 838 Amish subjects
Effect of high-fat meal on inflammatory markers
The distributions of basal WBC, CRP, IL-1β, MMP-1, and MMP-9 values were positively skewed. Postprandial changes in these markers were approximately normally distributed (see Supplementary Figure S1 online). Table 2 shows median levels (and interquartile ranges) of these markers at baseline (T0) and 4-h postchallenge (T4), as well as the mean changes (T4-T0) in inflammatory response to the high-fat meal intake. Levels of WBC, MMP-1, and MMP-9 were significantly elevated at T4 compared to T0, with increases of 16% for WBC count (P < 0.0001), 32.1% for MMP-1 (P < 0.0001), and 42.9% (P = 0.0009) for MMP-9. In contrast, median levels of IL-1β were essentially unchanged after the consumption of the high-fat meal. Triglyceride levels also increased significantly in response to the high-fat meal, from 68.1 mg/dl at T0 to 184.5 mg/dl at T4 (P < 0.0001). The increase in triglyceride levels was positively correlated with changes in WBC count (age- and sex-adjusted r = 0.12, P = 0.0007) and MMP-1 (age- and sex-adjusted r = 0.06, P = 0.06).
Table 2. Serum concentrations of inflammatory markers in response to a high-fat meal
Correlations among basal and postprandial inflammatory measures
The ln-transformed basal levels of WBC count, CRP, MMP-1, and MMP-9 levels were all significantly correlated with each other (r = 0.13–0.36, P < 0.0001 for all). The strongest age- and sex-adjusted residual correlations occurred between WBC count and MMP-9 (r = 0.36, P < 0.0001) and between WBC and CRP (r = 0.29, P < 0.0001). In contrast, basal IL-1β was significantly correlated only with basal levels of MMP-9 (r = 0.10, P = 0.007) and not with any of the other markers.
Additional analyses were carried out to determine relations between basal levels of WBC, MMP-1, and MMP-9 and the respective postprandial changes. Higher basal levels were significantly correlated with smaller postprandial changes in WBC (r = −0.14, P < 0.0001) and MMP-9 (r = −0.35, P < 0.0001) whereas there was no correlation between basal and postprandial MMP-1 change (r = −0.02, P = 0.5).
Predictors of basal levels of inflammatory levels
Age was positively correlated with higher basal levels of CRP (r = 0.34, P < 0.0001) and MMP-1 (r = 0.14, P < 0.0001). Basal MMP-9 levels were higher in men (age-adjusted geometric mean = 477.0 ng/ml and 422.2 ng/ml for men and women, respectively, P = 0.007) whereas basal CRP levels were higher in women (age-adjusted geometric mean = 0.8 mg/l and 1.0 mg/l for men and women, respectively, P = 0.003). Neither age nor sex was associated with WBC or IL-1β. The inflammatory markers were correlated with only a few covariates at levels greater than r = 0.2 (Table 3). In particular, higher BMI was significantly associated with higher basal CRP (r = 0.45, P < 0.0001) level, and higher fasting triglyceride was associated with higher basal WBC level (r = 0.21, P < 0.0001). In men, smokers had higher WBC count, CRP, and MMP-9 basal levels compared to nonsmokers. The age-adjusted geometric mean levels for smokers vs. nonsmokers were WBC = 5.7 vs. 5.0 thous/µl, CRP =1.0 vs. 0.7 mg/l, and MMP-9 = 551.8 vs. 454.8 ng/ml (P < 0.01 for all). Smoking was not significantly associated with IL-1β or MMP-1 basal levels.
Table 3. Correlations (r) between baseline characteristics and inflammatory markers at baseline and postprandial changes
Predictors of postprandial changes in inflammatory levels
Correlations of the baseline characteristics were generally much weaker with postprandial inflammatory changes than with basal inflammatory levels (Table 3). For example, postprandial changes in WBC count were correlated with only BMI, diastolic blood pressure, and fasting triglycerides, whereas postprandial changes in MMP-9 were correlated with only total cholesterol. None of these variables was significantly correlated with postprandial changes in MMP-1. In men, nonsmokers had a greater increases in WBC levels in response to fat intake than smokers in the age-adjusted model (age-adjusted WBC change = 0.76 thous/µl and 0.57 thous/µl, respectively; P = 0.04). In contrast, smoking status was not significantly associated with postprandial changes in either MMP-1 or MMP-9.
Genetic contributions to baseline and postprandial variation in inflammatory variables
Heritability of basal inflammatory factors (Table 4) ranged from 0.16 ± 0.08 for MMP-9 (P = 0.02) to 0.90 ± 0.07 for MMP-1 (P < 0.0001) in the age- and sex-adjusted model. Adjusting for covariates that were significantly correlated with the basal levels of each corresponding inflammatory markers did not change the results substantially. Heritabilities for postprandial inflammatory changes were much lower than those for basal levels. Only WBC change appeared to be modestly heritable (age- and sex-adjusted h2 = 0.14 ± 0.09, P = 0.04, and multivariable-adjusted h2 = 0.12 ± 0.09, P =0.07).
Table 4. Heritability estimates (h2 ± s.e.) of inflammatory markers
Consistent with previous studies suggesting that dietary fat may have important effects on inflammatory markers (8,9,10,11,12,13,14,15,24,28,29,30,31,32), we have shown significant increases in three: WBC count (16%), MMP-1 (32%), and MMP-9 (43%), following a standardized fat load. The postprandial increase in WBC count we observed is concordant with previous observations that blood leukocytes, particularly neutrophil counts, can increase within 3 h after fat (50 g/m2) intake in healthy normolipidemic young men (11,24). Our observation of postprandial increases in MMP-1 and MMP-9 are novel whereas the absence of significant postprandial increases in IL-1β in our data are consistent with previous report (28).
The determinants of basal levels and postprandial changes in inflammatory markers are largely unknown. Although some traditional cardiovascular risk factors (e.g., BMI, lipids, and blood pressure) are correlated with basal levels of some of these markers, the correlations with postprandial changes are substantially smaller. Consistent with earlier studies, we observed strong evidence for genetic contributions to basal inflammatory levels, with residual heritability estimates ranging from 15 to 89% after adjustment for the effects of other measured environmental covariates. Whereas strong genetic effects on serum CRP, WBC count, and IL-1β levels have been previously reported, few, if any, studies have provided heritability estimates for MMP-1 and MMP-9, two collagenases which affect atherosclerotic plaque stability via their ability to degrade extracellular matrix (33). In contrast to the strong genetic contributions to baseline levels, we detected only very weak evidence for genetic influences on postprandial inflammatory changes after fat intake; in fact, only for WBC count postprandial change was there evidence for even modest heritability (14 ± 9%; P = 0.04). It is possible that we were unable to detect stronger genetic effects on the postprandial variations in inflammation because the standardized fat challenge in this study was so strong that it overwhelmed the genetic control of the inflammatory response. Alternatively, despite the large number of family members studied, we may have had insufficient power to detect lower heritabilities (e.g., h2 < 20%). To our knowledge, heritabilities of postprandial changes in inflammatory markers have not been reported previously.
The increase in inflammatory activities in response to fat intake may result from elevated triglyceride-rich particles in the postprandial state. Hydrolysis of triglyceride in chylomicrons can change these particles into smaller remnant forms, which are hypothesized to be atherogenic because they can penetrate arterial tissue and accumulate within the subendothelial space (34). Chylomicron remnants as well as fatty acids released during lipoprotein lipase-mediated triglyceride hydrolysis are able to induce endothelial activation and dysfunction, expression of cellular adhesion molecules, and activation of monocytes and neutrophils, which can lead to the increase in the recruitment and activation of inflammatory cells to vascular endothelium and ultimately, the formation of foam cells (7,11,16,24,32,35). Postprandial triglyceride levels, particularly those measured 2–4 h after meal consumption, have been found to be more highly associated than fasting triglyceride levels with incident cardiovascular events in women, independent of levels of other lipids or traditional cardiovascular risk factors (23). Elevated postprandial triglycerides were also associated with increased common carotid intima-media thickness, suggesting that postprandial triglyceride may be crucial in the early stage of atherosclerosis (36,37). The significant positive correlations between postprandial triglyceride change and postprandial changes in WBC count and MMP-1 observed in this study support the hypothesis that inflammatory changes are related to postprandial triglycerides and may further imply a potential role of inflammation in early atherosclerosis.
The HAPI Heart Study has several unique features that are well suited for evaluating the determinants of the postprandial response of inflammatory factors to a high-fat challenge. These include a well-controlled intervention, the inclusion of large multiplex families and the uniformity of lifestyle and socioeconomic status, which serve to reduce nongenetic variability and boost the power to discern genetic determinants of traits. Nonetheless, several limitations need to be acknowledged. First, the high-meal challenge with 77.6% calories from fat represents a relatively extreme fat exposure such that the postprandial inflammatory changes observed in this study may not reflect the changes to a more moderate dietary fat intake. Second, the study did not include an external control group to evaluate the potential diurnal variations in inflammatory markers (38). To minimize the potential inflammatory variation over time, the high-fat meal intervention was performed in the morning after an overnight fast for all study participants. However, it is still possible that diurnal variation, to some extent, could influence the postprandial changes observed in this study. Lastly, the Amish lifestyle is characterized by a relatively high degree of physical activity relative to other populations (39). The degree to which physical activity modifies the effects of fat intake on inflammatory levels is unknown, although the relatively consistent associations between postprandial changes in inflammatory markers observed in this study and those reported from other studies (11,24) suggests that our finding that these changes are heritable in the Amish may be applicable also to other populations. It is also possible that the environmental homogeneity of the Amish lifestyle may boost our power to detect the genetic effects (heritability) on these traits.
In conclusion, our results provide strong evidence that inflammatory markers are under the influence of both modifiable risk factors, including exposure to dietary fat, and genetic factors. These findings suggest possible mechanisms linking genetic and environmental factors to inflammation and may ultimately provide insight into future CVD prevention. Additional studies are needed to identify loci that may functionally influence basal levels and to determine the relevance of the inflammatory response to a single high-fat meal.
Supplementary material is linked to the online version of the paper at http:www.nature.comoby
This work supported by NIH research grants U01 HL72515 and R01 088119, the University of Maryland General Clinical Research Center, grant M01 RR 16500, the Clinical Nutrition Research Unit of Maryland (P30 DK072488), and the Baltimore Veterans Administration Medical Center Geriatric Research and Education Clinical Center. We thank our Amish research volunteers for their long-standing partnership in research, and the research staff at the Amish Research Clinic for their hard work and dedication. Y.-C.C. was supported by the Merck Foundation fellowship.
The authors declared no conflict of interest.