点击显示 收起
1 From the Cardiovascular Nutrition Laboratory, Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, Boston, MA (NRM, SMJ, LMA, and AHL), and the Department of Medicine, Division of Cardiology, Tufts–New England Medical Center Hospitals, Tufts University School of Medicine, Boston, MA (JTK and RHK)
See corresponding editorial on page 927.
ABSTRACT
Background:The magnitude of the effect of soy protein on lipoprotein concentrations is variable. This discordance is likely attributable to the various forms of soy protein used and to unrecognized shifts in dietary fatty acid, cholesterol, and fiber.
Objective:The objective was to evaluate the effect of soybean processing as well as soy consumption relative to animal protein, independent of alterations in major dietary variables, on cardiovascular disease risk factors and vascular endothelial function.
Design:Twenty-eight hypercholesterolemic subjects (LDL cholesterol 3.36 mmol/L) aged >50 y consumed each of 4 diets for 6-wk periods according to a randomized crossover design. The diets [55% of energy as carbohydrate, 30% of energy as fat, and 15% of energy as protein—7.5% of energy as experimental protein (37.5 g/d)] were designed to contain products made from either whole soybeans, soyflour, or soymilk and were compared with a diet containing an equivalent amount of animal protein (meat, chicken, and dairy products). The cholesterol, fiber, and fatty acid profiles of the diets were equalized. All food and drink were provided, and body weight was maintained throughout the study.
Results:No significant differences in blood pressure, vascular endothelial function, or total cholesterol, VLDL-cholesterol, triacylglycerol, apolipoprotein B, or C-reactive protein concentrations were observed between the diets. Consumption of the soymilk diet resulted in a modest decrease (4%) in LDL-cholesterol concentrations compared with the animal-protein and soyflour diets (P < 0.05) and higher HDL-cholesterol (1%) and apolipoprotein A-I (2%) concentrations compared with the soybean and soyflour diets (P < 0.05).
Conclusions:The results suggest that the consumption of differently processed soy-based products and different types of protein (animal and soy) has little clinical effect on cardiovascular disease risk factors, including peripheral endothelial function, when other major dietary variables are held constant.
Key Words: Soy protein • lipids • lipoproteins • apolipoproteins • cholesterol • endothelial function
INTRODUCTION
Considerable interest has been focused on the role of soy protein and isoflavones in improving plasma lipoprotein profiles and subsequent cardiovascular disease (CVD) risk. This interest has stemmed from epidemiologic observations (1-4), data from early metabolic studies (5-10), and the results of a meta-analysis (11) showing that diets enriched in soy protein significantly reduced plasma cholesterol concentrations by 3–30% when compared with diets containing animal protein. In 1999, on the basis of the evidence from the available literature, the Food and Drug Administration approved a food label health claim of a reduced risk of heart disease from the consumption of 4 daily servings of foods containing 6.25 g soy protein or a total daily intake of 25 g/d (12).
Results from subsequent studies of soy protein and CVD risk have not been completely supportive of the original work, showing either a small or no plasma cholesterol reduction in normo- or hypercholesterolemic subjects (13-28). Interestingly, an early meta-analysis (11) concluded that the reduction in plasma cholesterol observed was not dependent on the amount of soy protein (8–124 g/d), but was strongly correlated with an individual's initial plasma cholesterol concentration. It is difficult to reconcile the earlier more pronounced effect of soy protein on plasma lipid concentrations with the subsequent data. One possible explanation is that a putative component of the soybean rather than of soy protein itself may be involved in the hypocholesterolemic effect and was unaccounted for in earlier studies. One such component studied that initially showed favorable effects on CVD risk factors was soybean isoflavones (4, 29, 30). However, results from more recent work do not support this hypothesis (28, 31, 32).
There are a variety of ways to process soybeans, and a large number of products are made from the different components. Although alcohol extraction and acid precipitation reduce the isoflavone content of the soy protein, other processing methods can alter soybean fiber, fat, sugars, phytic acid, and saponin content (33, 34). Consequently, the marked discordance in the data may be related to dissimilar forms of soy protein used in different studies. The aim of the present investigation was to assess the effects of differently processed forms of soybeans (whole soybean and products made thereof), of products derived from soyflour (textured soy protein), and of products made from soymilk (tofu and yogurt) between one another and relative to animal protein, independent of potential additional changes in the fatty acid, cholesterol, and fiber contents of the diet on CVD risk factors, including lipoprotein, apolipoprotein, and inflammatory markers in moderately hypercholesterolemic subjects. We also monitored peripheral vascular endothelial function as a surrogate marker of atherosclerosis and an indicator of vascular health and reactivity.
SUBJECTS AND METHODS
Subjects
Subjects who were 50 y of age with LDL-cholesterol concentrations 3.36 mmol/L at the time of screening, free of chronic illness, and not taking medications known to affect lipid metabolism (lipid-lowering drugs, ß-blockers, fish-oil capsules, cis-retinoic acid, ascorbic acid, vitamin E, diuretics, and hormones) were recruited for this study from the greater Boston area. Subjects reporting soy allergies, reporting the consumption of 2 alcoholic drinks/d, or who smoked were excluded from participation. All women were postmenopausal.
Twenty-one subjects initially recruited did not complete the study because of the following reasons: time constraints (n = 9), change in medical status (n = 3), noncompliance (n = 3), and gastrointestinal issues with diet foods (n = 6). With the exception of 2 subjects, all others terminated participation during phase 1 and were replaced. This high dropout rate was likely related to the long study duration (24-wk controlled diet period) and the relatively high level of daily soy protein (average intake: 37.5 g/d). Twenty-eight subjects (26 women and 2 men) completed the study and were included in the final statistical analysis. The characteristics of these study subjects (n = 28) at the time of screening are depicted in Table 1. The protocol was approved by the Human Investigation Review Committee of the Tufts University–New England Medical Center. All subjects gave written informed consents.
View this table:
TABLE 1. Baseline characteristics of the study participants1
Experimental design
The subjects were provided with each of 4 diets for 6-wk periods, separated by 2- to 4-wk intervening periods, according to a randomized crossover design. All food and drink were prepared by our metabolic research unit (MRU) and were provided to the subjects who reported to the MRU a minimum of 3 times/wk, at which time they were weighed, had their blood pressure recorded, and consumed one meal on site. The remaining food was packaged for takeout. The subjects were not allowed to supplement their diet with additional food or drink, with the exception of noncaloric beverages. The Harris-Benedict equation was used to calculate initial energy requirements of the subjects, and adjustments were made when necessary to maintain body weight.
Diets
The animal-protein diet was designed to be consistent with current dietary recommendations for CVD prevention (35) and provided 55% of energy as carbohydrate, 15% of energy as protein, and 30% of energy as fat (7% of energy as saturated fat, 10–15% of energy as monounsaturated fat, and 10% of energy as polyunsaturated fat), 80 mg cholesterol/1000 kcal, and 12-15 g fiber/1000 kcal. For the soy protein–based diets, 50% of the protein was contributed by soy-based foods. This was accomplished by displacing animal protein–containing products (meats, chicken, egg, and dairy products) with products made with whole soybean, soyflour, or soymilk. The whole-soybean phase contained whole organic soybeans (Westbrae Natural, Hain Food Group Inc, Uniondale, NY), soynuts (Solnuts Inc, Hudson, IA), defatted soyflakes (Cargill Inc, Cedar Rapids, IA), soya granules (Fearn Natural Foods, Mequon, WI), and soynut butter (Health Trip Co., Concord, MA). The soyflour phase contained soy nutlettes (Dixie USA Inc, Tomball, TX), whole-grain soyflour (Arrowhead Mills Inc, Hereford, TX), and textured soy-protein products (Chicken-Not, Beef-Not, Turkey-Not; Dixie USA Inc, Houston, TX). The soymilk phase contained tofu (Nasoya-Vitasoy USA Inc, San Francisco, CA), plain soy yogurt (Silk-Whitewave Food Co, Broomfield, CO), and plain soymilk (Silk-Whitewave Food Co). The nutrient profiles of the soy-based diets were balanced for carbohydrate, protein, fatty acid profile, and dietary cholesterol through the addition of butter, cream, and egg yolks so that the levels were equivalent to the animal protein–based diet. The soy products used to formulate the menus were chosen only from those commercially available in local grocery stores, supermarkets, and health food stores and had soy as the main ingredient. Provided in Table 2 are the macronutrient, fatty acid, cholesterol, fiber, oxalate, folate, and phytate contents (Covance Laboratories, Madison, WI); total isoflavone content, as aglycones (University of Iowa, Ames, IA); and total plant sterol concentrations (Cardiovascular Nutrition Laboratory, Tufts University, Boston, MA) of the diets. A 3-d rotating menu cycle was used during each phase, a sample of which is provided in Appendix A.
View this table:
TABLE 2. Composition of the 4 experimental diets1
View this table:
APPENDIX A. Protocol menu for the 4 diets1
Biochemical analyses
Three times during the sixth week of each of the 4 diet periods, blood samples were collected after a 14-h fast into EDTA-containing tubes. Plasma was separated by centrifugation at 1800 x g at 4 °C for 20 min. Total, HDL, and LDL cholesterol; triacylglycerol; and high-sensitivity C-reactive protein (CRP) concentrations were analyzed with a Hitachi 911 automated analyzer with the use of Roche Diagnostics reagents (Indianapolis, IN), and apolipoprotein (apo) A-I, and apo B were analyzed with Wako Diagnostics (Richmond, VA) reagents. VLDL cholesterol was calculated as the difference between total cholesterol and LDL plus HDL cholesterol. HDL and the subfractions were determined after selective precipitation with the use of a modified dextran sulfate–magnesium chloride method, and the cholesterol component was determined as previously described (22). Cholesterol assays were standardized through the Lipid Standardization Program of the Centers for Disease Control, Atlanta, GA.
Vascular endothelial function measurement
Peripheral vascular endothelial function was assessed with high-resolution ultrasound of the brachial artery during reactive hyperemia by standard methodology (36). Endothelial function measurements were performed at the end of each dietary phase by the same operator. Briefly, subjects were asked to lie down in a quiet, temperature-controlled, darkened room, and pulse rate and blood pressure were checked. Baseline measurements of the brachial artery in its longitudinal dimension were taken. Forearm ischemia was induced with a suprasystolic pressure cuff around the upper portion of the arm. After exactly 5 min of occlusion, reactive hyperemia ensued and the peak brachial artery diameter was measured between 45 and 75 s. Percentage flow-mediated dilation (FMD) was defined as the difference between the maximal brachial artery diameter during reactive hyperemia and the baseline brachial artery diameter divided by the baseline brachial artery diameter (%FMD = maximal brachial artery diameter – baseline brachial artery diameter/baseline brachial artery diameter).
Statistical analyses
The 3 values obtained per subject per variable were averaged, and this mean value was used in subsequent statistical analysis. Before the statistical testing was done, data were checked for normality, and appropriate transformations were performed when necessary (PROC UNIVARIATE; SAS version 9.1; SAS Institute Inc, Cary, NC). Variables log transformed included triacylglycerol, CRP, and total, VLDL-, HDL2-, and HDL3-cholesterol concentrations. An analysis of variance (PROC GLM) with main effect of diet and subject as repeated measures was carried out for each outcome measured, which was followed by a Tukey's honestly significant difference type of adjustment for the pairwise comparisons between each of the 4 treatment groups. Vascular endothelial function measurements were analyzed by using PROC GLM on rank-transformed data. Untransformed data are presented in the text and tables as means ± SD.
RESULTS
No significant differences in fasting plasma total, VLDL-cholesterol, triacylglycerol, apo B, or CRP concentrations were observed between the 4 diets (Table 3). LDL-cholesterol concentrations were lower after consumption of the soymilk diet than after consumption of the animal protein and soyflour diets (P < 0.05). HDL-cholesterol concentrations were higher after consumption of the animal-protein diet compared with the soyflour diet and higher after the soymilk diet compared with both the soybean and soyflour diets (P < 0.05 for all). These differences were attributable to changes in both the HDL2 and HDL3 subfractions. Apo A-I concentrations followed a pattern similar to that of HDL cholesterol, ie, no significant difference was observed between the animal-protein and soymilk diets, and concentrations were higher with the soymilk than with the soybean and soyflour diets (P < 0.05). The ratio of total cholesterol to HDL cholesterol was most favorable during the soymilk phase. No significant effect of the dietary perturbations in systolic or diastolic blood pressure or in peripheral endothelial function assessed by brachial artery FMD (Table 3) was observed. Stratifying and analyzing the data on the basis of LDL-cholesterol concentrations above (<3.80 mmol/L, n = 15) and below (3.80 mmol/L, n = 13) the median value and did not alter the conclusions (data not shown). Likewise, analysis of the data without the 2 male subjects did not change the conclusions.
View this table:
TABLE 3. Cardiovascular disease risk factors at the end of each dietary phase1
The percentage differences in LDL-cholesterol concentrations relative to the animal protein diet were –1.8%, 1.5%, and –3.9% and in HDL-cholesterol concentrations were –4.3%, –2.9%, and 0.7% after consumption of the soybean, soyflour, and soymilk diets, respectively. However, only the changes seen with the soymilk diet were statistically significant (Figure 1). No significant differences were seen in total cholesterol, triacylglycerol, and apo B concentrations (Figure 1) or in VLDL-cholesterol and CRP concentrations (data not shown).
FIGURE 1.. Mean (±SE) percentage difference in lipid, lipoprotein, and apolipoprotein profiles relative to the animal-protein diet. For each variable, bars with different lowercase letters are significantly different, P < 0.05 (PROC GLM and Tukey's honestly significant difference test). TC, total cholesterol; C, cholesterol; apo, apolipoprotein; TG, triacylglycerol.
To compare the effects of animal versus soy protein per se, data from the soybean, soyflour, and soymilk groups were combined to form the average Soy group and compared with the animal-protein diet. No significant effect of soy protein relative to animal protein was observed in plasma total, LDL-, HDL-cholesterol, triacylglycerol or total/HDL-cholesterol concentrations (Figure 2) or in the other CVD risk factors measured (data not shown).
FIGURE 2.. Mean (±SD) plasma lipid concentrations with the animal-protein (AP) and soy-protein (Soy) diets. The average response to the 3 Soy diets was compared with the response to the AP diet in a one-factor ANOVA at P < 0.05 (PROC GLM and Tukey's honestly significant difference test); no significant differences between groups were observed. TC, total cholesterol; C, cholesterol; TG, triacylglycerol.
DISCUSSION
Recent data suggest that the magnitude of the effect of soy protein on lipoprotein profiles is variable and less dramatic than originally reported (27, 28). This discrepancy may be due to the presence or absence of specific components associated with the soy-protein preparations or unaccounted shifts in the fatty acid, cholesterol, or fiber content of the diets used in earlier studies. The novel aspect of the present study was to directly compare the effect of soy protein derived from differently processed products, between each other and relative to animal protein, while simultaneously controlling for the macronutrient, fatty acid, cholesterol, and fiber contents of the diet. The results showed that the consumption of differently processed soy products or type of protein (animal protein compared with soy protein), when provided at approximately 1.5 times (37.5 g/d) the minimum amount on which the Food and Drug Administration–approved health claim was predicated had little clinical effect on CVD risk factors and vascular endothelial function.
The results of this intervention are consistent with the findings of 2 recent comprehensive reviews on soy and measures of cardiovascular health (28, 32). Both reports concluded that there was a significant, albeit modest, effect of soy protein with isoflavones when consumed at relatively high levels compared with milk or other proteins on LDL-cholesterol (–3% to –5%) but not on HDL-cholesterol or triacylglycerol concentrations. The evidence also favored soy protein rather than soy isoflavones as the responsible nutrient. However, the authors of these reports could not rule out the possibility of other soybean components as the active factor. We addressed this issue by comparing the effect of differently processed forms of soybean. Although the major dietary variables in our diets were equalized, isoflavone and trace components—such as plant sterols, folates, phytates, and oxalates—that could be altered by the processing method of the soy product and that have been shown or hypothesized to influence cholesterol concentrations were not equalized. Interestingly, processing appeared to have had little effect because the analysis of the final diets showed similar amounts of these components between diets (Table 2), which suggests that these components are not major contributing factors in the purported hypocholesterolemic effect of soy.
Our study also assessed the effect of soy consumption on other CVD risk factors: CRP concentrations, blood pressure, and vascular endothelial function. No significant differences were observed between diets. This is consistent with published studies evaluating the effect of soy consumption on CRP concentrations (18, 37, 38) and, with one exception (39), most studies on blood pressure (17, 18, 21, 24, 37). Similarly, most of the studies that assessed vascular endothelial function, either by increased brachial artery diameter (mm), FMD (%), flow (mL/min), or decreased peak flow velocity (cm/s), showed no effect or a slight improvement (18, 23, 40-43). Any beneficial effect was frequently attributed to the isoflavone component of the diet (23, 40, 41).
The reason for the more favorable plasma lipoprotein profile after consumption of the soymilk and not the soybean- and soyflour-enriched diets is unclear. Although no comparable studies were found in the literature, previous studies of soymilk have reported an average 5–8% reduction in plasma total cholesterol in normocholesterolemic subjects who consumed 350 mL soymilk/d (14.5 g soy protein) (43) and in hypercholesterolemic subjects who consumed 1 L soymilk/d (41.1 g soy protein) compared with cow milk (44). However, because of the lack of a control group fed an equivalent amount of protein (43), and dissimilar carbohydrate, unsaturated fat, fiber, and cholesterol contents of the diets (44), the hypocholesterolemic effect observed in these studies cannot be solely attributed to the soymilk supplementation.
It has been hypothesized that the specific amino acid profile of the soy protein may be the active agent found in soy (45). Data from animal experiments (46, 47) have shown that the amino acids lysine and methionine tended to be hypercholesterolemic, whereas arginine was hypocholesterolemic, which suggests that the higher arginine to lysine and methionine amino acid profile of soy protein might explain, at least in part, its hypocholesterolemic effects. In the present study, despite the use of a variety of soy products, the amino acid profiles of the 3 soy-based diets were similar (ratios of arginine to lysine of 0.8, 1.1, 1.2, and 1.1 for the animal-protein, soybean, soyflour, and soymilk diets, respectively). Whether the incorporation of even higher amounts of the soy-based products would have sufficiently altered the amino acid profile to elicit a greater hypocholesterolemic response than observed is unknown. More recent studies (48, 49) have shown that the 7S globulin protein present in soy can up-regulate LDL receptors in vitro. The authors suggest that this mechanism might explain the discordance in results because the concentration of the 7S globulin differs among soy varieties and, consequently, in products prepared thereof. Insufficient data exist to draw conclusions regarding this postulated mechanism in the present study.
A limitation of the present work was the use of a high amount of protein (7.5% of total protein intake) from a single source of soy and its products, in contrast with the more likely real life situation in which multiple animal and vegetable protein sources are habitually consumed. However, this approach maximized our ability to detect any potential beneficial effect were one present. Another limitation is that we did not directly test the effect of substituting each category of soy products for animal protein–based products without controlling for the multiple shifts in associated dietary variables. To do that would not have allowed us to address our primary hypothesis, partitioning out the independent effect of the soy-protein source, per se, on CVD risk factors. For similar reasons we did not attempt to obtain isolated subfractions of individual soybean components to assess the effect of each on the outcome measures. A strength of this study was that soy protein from differently processed but readily available products, and not from one type of soy protein isolate or specialized preparation of soy protein, was assessed relative to each other and animal protein. Another strength of the study was that moderately hypercholesterolemic subjects served as study subjects, the population most likely to benefit from alterations in the diet to improve CVD risk factors.
In conclusion, relatively high intakes of differently processed forms of soy protein derived from commonly available soy products appear to have little effect on CVD risk factors, either between each other or when substituted for animal protein. Nonetheless, the use of foods containing soy protein to displace animal protein, which would result in reductions in saturated fat and cholesterol intakes, could be beneficial to overall CVD risk reduction.
ACKNOWLEDGMENTS
We are indebted to the Metabolic Research Unit staff (the study volunteer recruiters, the nurses for their expert care of the study subjects, and the dietitians and kitchen staff for their meticulous planning and preparation of the study foods for each dietary phase), to Kathleen Sliney and Paula Heggericks for technical assistance with the vascular endothelial function measurements, and to the study subjects without whom this investigation would not have been possible.
AHL (principal investigator) and NRM (project coordinator) designed, implemented, and oversaw all aspects of this study. NRM wrote the initial draft of the manuscript. AHL, SMJ, LMA, JTK, and RHK contributed to critically reviewing the manuscript. SMJ was responsible for the biochemical analysis. JTK and RHK were responsible for supervising the vascular endothelial measurements. LMA was responsible for the statistical analysis. None of the authors had any conflicts of interest.
REFERENCES
Related articles in AJCN: