Baseline Serum C-Reactive Protein Is Associated with Lipid Responses to Low-Fat and High-Polyunsaturated Fat Diets^,

Abstract

Baseline serum C-reactive protein (CRP) concentrations play a role in the lipid response to diet. This study was a randomized, cross-over, controlled feeding study with 3 phases of 25 d each aimed at determining whether baseline CRP concentrations modulate the serum lipid response to diets differing in fat type and quantity. Participants were adult men and women, age 19–65 y, with LDL-cholesterol concentrations of 3.37–4.66 mmol/L. All participants consumed 3 diets differing in the type of snack, either low or high in fat: low-fat (30.8% of energy), moderate in fat and saturated fat (37.9 and 11.4% of energy, respectively), or moderate in fat and polyunsaturated fat (36.3 and 9.7% of energy, respectively). Using baseline CRP as a continuous variable, CRP × diet interactions on change in serum lipoprotein_a (P = 0.045) and HDL-cholesterol (P = 0.06) were observed. When we used previously established categories to define CRP concentrations (low, <1 mg/L; intermediate, 1–3 mg/L; or high, >3 mg/L), we found a CRP × diet interaction on change in triglyceride concentrations (P = 0.03) and trends for CRP × diet interaction on change in LDL (P = 0.06) and total cholesterol (P = 0.07). If replicated, these results suggest that considering baseline CRP concentrations when prescribing dietary interventions to lower lipid concentrations may be useful. Individuals with high baseline CRP concentrations may benefit from moderate-fat, high polyunsaturated fat diets, whereas those with low baseline CRP concentrations may obtain greater lipid-lowering benefits from low-fat diets.

Introduction

In 2003, the AHA and the CDC Scientific Statement on Inflammatory Markers recognized high sensitivity C-reactive protein (hs-CRP)⁷ as a useful marker of cardiovascular disease (CVD)⁷ risk (1). Specifically, the authors highlighted the usefulness of hs-CRP as an additional risk predictor for supporting additional lifestyle or drug therapies in patients at risk of CVD (1). Recently, studies have been published proposing a role for hs-CRP in determining lipid responses to various dietary interventions (2,3). These studies suggest that lipid responses to dietary interventions may differ based on a person's baseline CRP concentrations.

For example, in a study by Desroches et al. (2), men with low baseline CRP concentrations had reduced fasting triglycerides (TG) after consuming a low-fat (LF) diet, whereas those with high CRP at baseline had increased TG after consuming a LF diet. Hilpert et al. (3) also found differences in lipid responses to dietary interventions depending on baseline CRP concentrations; men with low CRP at baseline had decreased LDL-cholesterol after consuming a LF diet, whereas those with high baseline CRP had increased LDL-cholesterol after consuming the same diet. Data from a high monounsaturated fatty acid (MUFA) diet showed that men with high baseline CRP had decreases in total cholesterol (TC) and LDL-cholesterol that were greater than reductions in men with low baseline CRP (2). These studies suggest that lipid responses to LF and high-MUFA diets differ based on an individual's baseline CRP concentrations.

We recently conducted a study in which we assessed the impact of substituting snacks differing in fat content on CVD risk (4). Our diets were either LF or moderate in fat, with the 2 moderate fat diets differing mostly in their PUFA and saturated fat content. We used data from this study to determine whether baseline CRP concentrations could predict the lipid response to the different diets.

Methods

Participants and study design.

Data for the analyses reported here were obtained as part of a previous study. Details of the study participants and protocol are reported elsewhere (4). Briefly, a total of 45 participants (n = 12 men and 33 women) aged between 19 and 65 y were enrolled in the main study. All participants had fasting serum LDL-cholesterol concentrations between 3.37 and 4.66 mmol/L, TG <3.96 mmol/L, and glucose concentrations <6.99 mmol/L, and BMI between 20 and 35 kg/m². Participants were weight stable and nonsmokers. We instructed participants to maintain their regular level of physical activity and not to change their medication use throughout the study. Characteristics of the participants by baseline CRP category are shown (Table 1).

TABLE 1.

Participant characteristics by baseline CRP category¹

Characteristic	Low CRP2	Intermediate CRP	High CRP
Age, y	42.6 ± 8.7	41.3 ± 13.8	46.9 ± 11.3
Weight, kg	80.1 ± 11.4	83.5 ± 19.7	80.3 ± 8.7
BMI, kg/m2	29.5 ± 5.0	29.5 ± 6.9	30.0 ± 3.3
Waist circumference, cm	89.6 ± 8.4	97.6 ± 13.8	100.0 ± 10.7
Serum TC, mmol/L	5.80 ± 0.23	5.79 ± 0.74	5.61 ± 0.70
Serum LDL-cholesterol, mmol/L	4.06 ± 0.30	3.85 ± 0.62	3.63 ± 0.51
Serum HDL-cholesterol, mmol/L	1.25 ± 0.085	1.30 ± 0.19	1.46 ± 0.29
Serum TG, mmol/L	1.07 ± 0.22	1.72 ± 1.09	1.30 ± 0.61

¹Data are means ± SD, n = 33.

²Low CRP, < 1mg/L; intermediate CRP, 1–3 mg/L; high CRP, > 3 mg/L.

The study design was a randomized cross-over trial with 3 controlled feeding phases of 25 d each separated by washout periods of 4 or 8 wk. The study was conducted at the General Clinical Research Center of the University of Alabama at Birmingham (Birmingham, AL) and diets were prepared by the Bionutrition Unit. The University of Alabama at Birmingham Institutional Review Board Study approved the study protocol and procedures and all participants signed informed consent forms prior to the start of the study.

Study diets.

The 3 diets were designed based on a control, National Cholesterol Education Program Step 1 diet. The control diet (LF) contained low-fat snacks and the other 2 diets contained higher fat snacks, which differed in fatty acid profile: either high in saturated and trans fat [Western diet (WD)] or high in (n-6) PUFA (HPUFA). These moderate-fat diets provided 36.3 and 37.9% of energy from fat (HPUFA and WD, respectively). The macronutrient breakdown of the diets is shown (Table 2). The base diet was identical among the 3 diets and differed only in the types of snacks included. The main PUFA in the diets were of the (n-6) series. Snacks provided ∼12–15% of energy requirements (1255 kJ/d), depending on the weight maintenance energy prescription of each participant. We calculated the weight maintenance energy requirements using the Harris-Benedict equation with an activity factor of 1.35 (5). Adjustments to the energy prescription were made after d 7 of the study if weight was altered by >1%.

TABLE 2.

Macronutrient content of the test diets

Nutrient	LF diet	HPUFA diet	WD
	%
Carbohydrate	54.8	48.6	46.0
Protein	14.7	15.5	16.3
Fat	30.8	36.3	37.9
PUFA	5.2	9.7	5.8
MUFA	14.2	15.3	15.9
SFA	8.5	8.5	11.4
Trans fatty acids	1.2	0.7	2.7

All meals were prepared by the General Clinical Research Center and breakfast was consumed under supervision each weekday morning. All other meals were packaged for home consumption. Participants were required to consume all foods and beverages provided and nothing else, except for nonnutritive beverages and condiments. Each weekday morning, participants provided daily report forms to report any symptoms of illness, foods not eaten, beverages consumed, and medication(s) taken. If participants did not consume a study food on a given day, they were instructed to consume the food the following day. Compliance with the study was very good and participants did not report deviating from protocol.

Clinical measurements.

We collected blood samples from fasting participants at baseline, d 15, and d 25 of each phase. Blood samples were analyzed for complete lipoprotein profile using the vertical auto-profile technique (6) (Atherotech). Serum concentrations of TC, HDL-cholesterol, including subparticles HDL₂ and HDL₃, LDL-cholesterol, VLDL cholesterol, and lipoprotein a (Lp_a), as well as LDL particle density, were assessed. We measured hs-CRP concentrations on an ILAB 600 automatic analyzer using test kits (Instrumentation Laboratories).

Statistical analyses.

We analyzed the data using mixed-model ANOVA with unstructured covariance matrices. The dependent variables were percent changes in lipid measurements from the average of d 15 and d 25 to baseline relative to the WD diet. Independent variables included age, gender, time, phase, diet, and baseline CRP. The primary effect of interest was a diet × baseline CRP interaction. All analyses were conducted using CRP as a continuous variable and then repeated with CRP as a categorical variable. In the categorical analysis, participants were grouped into low, intermediate, and high CRP concentrations if their baseline CRP concentration was <1, 1–3, or >3 mg/L, according to the classification system established by the AHA and the CDC (1). An interaction term with a P-value of < 0.05 was considered significant. Data are presented as means ± SEM.

Results

A total of 33 participants completed all 3 phases of the study (n = 7 men and 26 women). Participant characteristics based on their CRP concentration at study entry (phase I, d 1) are shown (Table 1). Overall, diet did not affect hs-CRP concentrations.

CRP as a categorical variable.

When data were analyzed using CRP as a categorical variable, there was no diet × baseline CRP interaction on HDL-cholesterol (P = 0.18), Lp_a (P = 0.44), or risk of having LDL particle type B (P = 0.85). There was a nearly significant interaction effect on LDL-cholesterol (P = 0.06) and TC (P = 0.07) and a significant interaction effect on TG (P = 0.03; Table 3). The HPUFA diet tended to reduce LDL-cholesterol relative to the WD diet in all CRP groups, whereas the LF diet reduced LDL-cholesterol to a greater extent in the low and intermediate CRP groups but not in the high CRP group. A similar nearly significant interaction was found for TC, except that there was a step-wise incremental TC reduction with increasing CRP category on the HPUFA diet. TG concentrations were reduced to the same extent with the HPUFA and LF diets relative to the WD in participants with low baseline CRP concentrations. In participants with intermediate and high CRP concentrations at baseline, TG increased with the consumption of the LF diet relative to the WD, whereas for the high CRP group, TG decreased with consumption of the HPUFA diet relative to the WD.

TABLE 3.

Percent change in serum lipid concentrations relative to the high-fat diet in individuals with low, intermediate, and high CRP concentrations at baseline: categorical CRP values¹

	LDL-cholesterol		TC		HDL-cholesterol		TG		Lp_a
Diet	LF	HPUFA	LF	HPUFA	LF	HPUFA	LF	HPUFA	LF	HPUFA
Low CRP2	−9.20 ± 0.27	−2.92 ± 0.23	−6.27 ± 0.22	−1.32 ± 0.19	1.56 ± 0.28	4.62 ± 0.24	−29.06 ± 0.84	−28.86 ± 0.71	5.95 ± 1.83	32.28 ± 1.57
Intermediate CRP	−4.30 ± 0.17	−1.95 ± 0.17	−5.27 ± 0.15	−2.69 ± 0.14	−6.58 ± 0.21	0.30 ± 0.18	7.47 ± 0.59	−0.16 ± 0.55	−35.88 ± 1.36	8.57 ± 1.28
High CRP	−0.10 ± 0.13	−4.19 ± 0.13	0.04 ± 0.10	−3.06 ± 0.11	2.17 ± 0.13	2.63 ± 0.14	12.31 ± 0.40*	−11.74 ± 0.40	−16.83 ± 0.87	−4.41 ± 0.92

¹Values are means ± SEM, based on the fitted model. *Different from HPUFA, P < 0.05.

²Low CRP, < 1 mg/L; intermediate CRP, 1–3 mg/L; high CRP, > 3 mg/L.

CRP as a continuous variable.

When data were analyzed using CRP as a continuous variable, there was no diet × baseline CRP interaction on LDL-cholesterol (P = 0.45), TC (P = 0.17), TG (P = 0.69), or risk of having LDL particle type B (P = 0.97). There was a nearly significant interaction effect on HDL-cholesterol (P = 0.06) and a significant interaction effect on Lp_a (P = 0.04; Table 4). For HDL-cholesterol, the HPUFA diet tended to increase HDL-cholesterol relative to the WD for all participants, whereas the LF diet tended to reduce HDL-cholesterol relative to the WD in the low and intermediate CRP groups. Participants with high baseline CRP concentrations had no change in HDL-cholesterol concentrations with consumption of the LF diet relative to a WD. For Lp_a, there was an increase relative to the WD with consumption of the HPUFA diet in individuals with all CRP groups and reductions in Lp_a relative to the WD with consumption of the LF diet in all individuals. Participants with low and intermediate baseline CRP had greater reductions in Lp_a than those with high baseline CRP with consumption of the LF diet relative to the high-fat diet.

TABLE 4.

Percent change in serum lipid concentrations relative to the high-fat diet in individuals with low, intermediate, and high CRP concentrations at baseline: continuous CRP values¹

	LDL-cholesterol		TC		HDL-cholesterol		TG		Lp_a
Diet	LF	HPUFA	LF	HPUFA	LF	HPUFA	LF	HPUFA	LF	HPUFA
Low CRP2	−5.11 ± 0.15	−4.80 ± 0.14	−5.03 ± 0.12	−3.45 ± 0.11	−3.61 ± 0.14	2.68 ± 0.13	2.95 ± 0.45	−9.33 ± 0.44	−35.08 ± 0.94	12.36 ± 0.86
Intermediate CRP	−4.57 ± 0.12	−4.47 ± 0.12	−4.38 ± 0.10	−3.17 ± 0.10	−2.89 ± 0.12	2.60 ± 0.12	3.78 ± 0.39	−9.34 ± 0.38	−30.31 ± 0.80	11.53 ± 0.74
High CRP	−2.53 ± 0.10	−3.27 ± 0.10	−1.96 ± 0.09	−2.13 ± 0.08	−0.20 ± 0.10	2.33 ± 0.10	6.85 ± 0.32*	−9.37 ± 0.33	−12.47 ± 0.65	8.40 ± 0.62

¹Values are means ± SEM, based on the fitted model. *Different from HPUFA, P < 0.05.

²Low CRP, < 1 mg/L; intermediate CRP, 1–3 mg/L; high CRP, > 3 mg/L.

Discussion

This study shows that individuals differing in baseline CRP concentrations have different lipid responses to small differences in the fatty acid profile of their diet. In this study, test diets either low or moderate in fat and either low or moderate in saturated and polyunsaturated fats were consumed at a level required to maintain weight. Individuals with low baseline CRP concentrations tended to have greater reductions in TC and LDL-cholesterol with the consumption of the LF diet compared with the moderate-fat, high-saturated fat WD, whereas those with high baseline CRP tended to have greater reductions in TC and LDL-cholesterol with the consumption of the moderate-fat, HPUFA diet compared with consumption of the moderate-fat, high-saturated fat WD. In addition, individuals with high baseline CRP showed a significant reduction in TG relative to the WD with consumption of the HPUFA diet that significantly differed from the increase in TG relative to the WD with consumption of the LF diet. A change from a moderate-fat, high saturated fat diet to a moderate-fat, HPUFA diet may therefore be preferentially advocated for individuals with high CRP concentrations.

Our data showing that a LF diet decreases TC and LDL-cholesterol relative to a high-fat diet in individuals with low baseline CRP concentrations are similar to those observed by Erlinger et al. (7). In that study, participants consumed a low-fat Dietary Approaches to Stop Hypertension (DASH) diet or a control, high-fat diet for periods of 30 d within a controlled feeding condition. Participants were categorized as having low or high CRP concentrations based on being below or above the median CRP concentration for that sample (2.8 mg/L for the control diet and 1.7 mg/L for the DASH diet). Individuals below the median for CRP had significant reductions in TC and LDL-cholesterol with DASH diet consumption relative to the control diet. Similarly, Hilpert et al. (3) found that participants with low baseline CRP (below their sample 3.5 mg/L median) had significant reductions in LDL-cholesterol and the LDL:HDL-cholesterol ratio with consumption of an AHA Step I diet. On the other hand, Desroches et al. (2) found no diet × CRP interaction on TC and LDL-cholesterol in men consuming either a low-fat AHA Step II diet or a high-fat, high MUFA diet. These conflicting results may be due to gender differences between studies; the study by Desroches et al. (2) was conducted on men only, whereas our study and those of Erlinger et al. (7) and Hilpert et al. (3) were conducted in a sample of ∼≥50% women. Gender differences in the effects of CRP concentrations on lipid responses to diets have not been reported previously.

In this study, we found reductions in TC and LDL-cholesterol with consumption of a moderate-fat, HPUFA diet relative to a high-fat, high-saturated fat diet in individuals with high baseline CRP concentrations. Only 1 previous study specifically examined the impact of CRP concentrations on lipid responses to a high-PUFA diet (8). In this study, individuals with low CRP concentrations at baseline (<2 mg/L) had significant reductions in LDL-cholesterol on high-PUFA diets rich in linoleic acid or α-linolenic acid relative to a high-fat, average American diet. In our study, participants with high baseline CRP had greater reductions in LDL-cholesterol after consuming a high-PUFA diet. Differences in study results, if replicable, may be due to differences in PUFA content of the diets: ∼10% of energy in our study vs. 12–13% of energy in the study by Zhao et al. (8). Differences may also be due to the PUFA series of the diets; our diets differed mostly in (n-6) PUFA content, whereas diets in Zhao et al. (8) differed mostly in (n-3) PUFA content. However, our results are analogous to those obtained with a high-MUFA diet. In the study by Desroches et al. (2), men with high baseline CRP concentrations had significant reductions in TC and LDL-cholesterol after consuming a high-fat, high-MUFA diet. Higher fat diets rich in unsaturated fats may therefore be beneficial in lowering cholesterol concentrations in those individuals with high CRP concentrations.

Our results are also similar to those of Desroches et al. (2) and Erlinger et al. (7) regarding TG concentrations. In our study and that of Desroches et al. (2), individuals with low baseline CRP had reduced TG after consuming a LF diet, whereas in all 3 studies, those with high baseline CRP had increased TG concentrations after a LF diet. However, unlike our data, Desroches et al. (2) did not find greater reductions in TG with the consumption of a high-fat, high MUFA diet in those with high CRP concentrations. In the present study, the change in TG with a moderate-fat diet high in PUFA relative to a moderate-fat diet higher in saturated fat was significantly different from the change in TG with a LF diet relative to a moderate-fat diet higher in saturated fat. It may be that PUFA may have a better TG-lowering effect in those individuals with high CRP concentrations than MUFA. Studies aimed at directly comparing MUFA- and PUFA-rich diets would be necessary to determine whether this is the case. However, in general, MUFA and PUFA diets tend to exert similar effects on TG concentrations (9).

Our study has a few limitations. First, our primary objective was not to test CRP's modulation of lipid responses to different fat intakes. Therefore, we did not plan to have equal numbers of individuals in each CRP category, nor did we match our participants. Perhaps with larger samples, we would have reached significance for CRP by diet interactions on TC and LDL-cholesterol and closer agreement between categorical and continuous CRP analyses. Furthermore, in this study, we chose to conduct our analyses using CRP as both a continuous variable and as a categorical variable. Using CRP as a categorical variable allows us to directly compare our data with previous studies examining a similar question. However, each study differs in the manner in which CRP is categorized. We chose to use the AHA/CDC values for low, intermediate, and high CRP (1). Others have used the median value of their sample population to distinguish between low and high CRP concentrations (3,7,8) or used 1 AHA/CDC cutpoint to provide 2 categories, low and high CRP (2). Nevertheless, data are consistent between studies. Using CRP as a continuous variable has the advantages that it does not degrade the data, does not require the selection of arbitrary cutpoints, tends to preserve statistical power, and can make the interpretation of interaction effects easier (10–15). By using both analyses, we could evaluate the sensitivity of the results to the analytic strategy. Indeed, the results did differ by analytic strategy, suggesting that they warrant confirmation in future research and should be considered as hypothesis generating.

Another limitation of this study is the lack of measurement of other inflammatory markers or enzymes involved in lipoprotein metabolism, such as interleukin-6, tumor necrosis factor-α, or lipoprotein lipase. Interleukin-6 is secreted in response to interleukin-1 and tumor necrosis factor-α and leads to stimulation of lipolysis (16,17). In addition, interleukin-6 leads to increases in CRP concentrations (16). High concentrations of CRP could selectively bind to LDL particles, leading to increased complement-activating capacity and progression of atherosclerosis (18). Our data suggest that dietary factors such as quantity or quality of fat in the diet have a greater impact on plasma lipids in different inflammatory states. Yet, how the inflammatory status of an individual modulates his or her lipid responses to diet remains to be determined.

Finally, our data add to the body of literature showing that a person's inflammatory state, reflected by CRP concentrations, can predict his/her lipid responses to dietary manipulations. Specifically, individuals with high CRP concentrations would benefit from the consumption of a moderate-fat, high-PUFA diet, whereas individuals with low CRP concentrations may show greater benefit from consumption of a low-fat diet. Our data show that improvements in lipids based on baseline CRP concentrations can be achieved with simple whole-foods substitutions.