| HOME | Previous Columns | Email Contacts | Advertising |
Trans unsaturated fatty acids are produced commercially in large quantities by heating vegetable oils in the presence of metal catalysts and hydrogen to form shortening and margarine. They are so named because the carbon atoms adjacent to their double bonds are on opposite sides, resulting in a straight configuration and a solid state at room temperature. In contrast, naturally occurring unsaturated fatty acids contain double bonds as cis isomers, with adjacent carbons on the same side of the double bond, resulting in a bent shape and a liquid state at room temperature. Partial hydrogenation, the process used to create trans fatty acids, is primarily used to produce solid fats. However, it also removes essential polyunsaturated fatty acids, such as linolenic acid and linoleic acid, because they tend to oxidize, causing the fat to become rancid with prolonged storage or when exposed to the high temperatures used for commercial deep-fat frying. Trans fatty acids are also produced in the rumen of cattle, resulting in low levels of these isomers in dairy and beef fat.
Production of partially hydrogenated fats began early in the 20th century and increased steadily until about the 1960s, as processed vegetable fats displaced animal fats in the diets of most people in industrialized countries. The initial motivation was lower cost, but health benefits were later purported. Levels of trans fatty acids in margarines have declined as softer margarines have become popular. The average per capita consumption of trans fatty acids from partially hydrogenated oils has remained at about 2 percent of calories since the 1960s, because of the increased use of these fats in commercially baked products and fast foods.
By the early 1990s it became apparent that the consumption of trans fatty acids had uniquely adverse effects on blood lipid levels in metabolic studies and was associated with an increased risk of coronary heart disease in epidemiologic investigations. A 1995 industry-sponsored review concluded that there was insufficient evidence to take action and that further research was needed. Since then many more metabolic and epidemiologic studies have confirmed the adverse effects of trans fatty acids, stimulating the Food and Drug Administration to announce plans to include the trans-fatty-acid content of foods on product labels. One important issue is whether to list the amount of trans fatty acids separately or to combine it with the saturated-fat content. In this article we shall review the effects of trans fatty acids on blood lipid levels that have been identified in metabolic studies and the associated risk of coronary heart disease that has been identified in epidemiologic studies.
Early metabolic studies generally found that the cholesterol-raising effect of hydrogenated fat was less than that of saturated fats. However, the focus on total cholesterol masked the fact that although trans fatty acids and saturated fatty acids increase low-density lipoprotein (LDL) cholesterol levels to a similar degree, trans fatty acids also lower high-density lipoprotein (HDL) cholesterol levels. A 1990 study demonstrated that the replacement of a diet high in oleic acid (10 percent of the daily energy intake), the primary monounsaturated fat in diets, with a diet high in trans fatty acids increased LDL cholesterol levels by 14 mg per deciliter (0.37 mmol per liter) and decreased HDL cholesterol levels by 7 mg per deciliter (0.17 mmol per liter). In contrast, replacement of oleic acid with saturated fatty acids caused a similar increase in LDL cholesterol levels, but had no effect on HDL cholesterol levels. As a result, the ratio of LDL cholesterol to HDL cholesterol was significantly higher with the trans-fatty-acid diet (2.58) than with the saturated-fat diet (2.34) or the oleic-acid diet (2.02). These findings have been confirmed in many studies, including the study by Lichtenstein et al. reported in this issue of the Journal, with the use of various levels and mixtures of trans fatty acids. summarizes the randomized trials that directly compared the effects of trans fatty acids with those of isocaloric amounts of cis fatty acids.When the data are available, the figure also shows the effects of saturated fatty acids in the same studies.
Because trans fatty acids increase LDL cholesterol to levels similar to those produced by saturated fatty acids and also decrease HDL cholesterol levels, the net effect of trans fatty acids on the ratio of LDL cholesterol to HDL cholesterol is approximately double that of saturated fatty acids. The only somewhat discordant result was from a small Malaysian study, which found a considerably stronger adverse effect of trans fatty acids; we have conservatively excluded this result in estimating the regression line in We also did not include the study by Almendingen et al. in because they did not compare a diet high in trans fatty acids with a diet high in oleic acid or polyunsaturated fat. Almendingen et al. found that trans fatty acids from hydrogenated fish oil but not from hydrogenated soybean oil increased the ratio of LDL cholesterol to HDL cholesterol more than did butter.
The effect of trans fatty acids on the ratio of LDL cholesterol to HDL cholesterol was significantly larger than that of saturated fatty acids in each of the six studies that allowed a direct comparison. Collectively, these studies provide definitive evidence that trans fatty acids raise this ratio more than do saturated fatty acids. As shown by the best-fit regression line in , an absolute increase of 2 percent in the intake of trans fatty acids would raise the ratio of LDL cholesterol to HDL cholesterol by 0.1 unit. Since a 1-unit increase in the ratio is associated with a 53 percent increase in the risk of coronary heart disease, the average intake of 2 percent of calories from trans fatty acids in the United States would be predicted to account for a substantial number of deaths from coronary heart disease.
Other trials have compared the effects of butter and margarine on blood lipid levels. Because margarines are usually higher in cis fatty acids than butter, the specific effects of trans fatty acids cannot be estimated accurately from these trials. A meta-analysis of these investigations, however, showed that butter and stick margarines, which typically contain 20 to 25 percent trans fatty acids, have similar effects on the ratio of total cholesterol to HDL cholesterol, whereas soft margarines, which are low in trans fatty acids, lower the ratio. These results confirm the deleterious effects of trans fatty acids on blood lipid levels and indicate that these may offset the beneficial effects of polyunsaturated fat.
Besides increasing the ratio of LDL cholesterol to HDL cholesterol, trans fatty acids increase Lp(a) lipoprotein levels when they are substituted for saturated fatty acids. A significant increase in Lp(a) lipoprotein levels was reported in 9 of 10 trials, with an average increase of 0.5 mg per deciliter per 2 percent of energy intake from trans fatty acids. High blood levels of Lp(a) lipoprotein have been associated in some studies with an increased risk of coronary heart disease; the effect of the smaller variations in blood levels of Lp(a) lipoprotein induced by trans fatty acids is uncertain.
Trans fatty acids also raise triglyceride levels measured while subjects are fasting. Numerous studies have reported increases in triglyceride levels ranging from 1.0 to 24 mg per deciliter (0.01 to 0.27 mmol per liter), with an average increase of 3.0 mg per deciliter (0.03 mmol per liter) per 2 percent of energy intake from trans fatty acids.The effect of such an increase on the risk of cardiovascular disease, though limited, is probably unfavorable.
The strong correlation between the level of intake of saturated fatty acids and the rates of coronary heart disease among the 16 populations examined in the Seven Countries Study is often quoted as evidence that the consumption of saturated fat increases the risk of coronary heart disease. A subsequent biochemical analysis of food composites representing the average intake of each cohort at base line not only confirmed that the intake of saturated fatty acids was strongly correlated with the risk of death from coronary heart disease (r=0.88, P<0.001) but also showed that the intake of trans fatty acids was correlated with the risk of death from coronary heart disease (r=0.78, P<0.001). Interpretation of comparisons among populations with widely different lifestyles is hazardous, but these data leave room for a potentially substantial effect of trans fatty acids on the risk of coronary heart disease.
Several case-control or cross-sectional studies have also been conducted. In a case-control study of subjects in the Boston area, we found a strong and significant positive association between the intake of trans fatty acids, assessed with the use of dietary questionnaires, and the risk of acute myocardial infarction. The relative risk of acute myocardial infarction for the quintile with the highest intake of trans fatty acids as compared with the quintile with the lowest intake was 2.4 (P for trend <0.001); this association was entirely explained by the intake of these fats from hydrogenated vegetable oil. Bolton-Smith et al. performed a cross-sectional analysis of the association between the intake of trans fatty acids and the presence of previously undiagnosed coronary heart disease among participants in the Scottish Heart Study. The intake of trans fatty acids was positively correlated with the ratio of LDL plus very-low-density lipoprotein cholesterol to HDL cholesterol. The odds ratios for coronary heart disease in the quintile with the highest intake as compared with the quintile with the lowest intake were elevated but not significantly so (1.26 in women and 1.08 in men).
Studies in which the composition of fatty acids in tissue or plasma was used as a marker of trans-fatty-acid intake have yielded conflicting results. With one exception, however, these studies have been too small to detect an association reliably. The results of the only large study, which included 671 men with acute myocardial infarction from eight European countries, were inconclusive. The overall analyses revealed no association between the intake of trans fatty acids and the risk of myocardial infarction. However, in contrast to the centers studied in other countries, the two centers studied in Spain, where the rates of coronary heart disease are very low, reported extremely low levels of trans-fatty-acid intake and little variation between subjects and thus provided little information. After the exclusion of these data, the odds ratios for the third and fourth quartiles of intake, as compared with the lowest, were 1.53 and 1.44, respectively. The interpretation of the results of this study has caused controversy, but in any case they do not provide strong evidence against the hypothesis that the consumption of trans fatty acids increases the risk of coronary heart disease.
The strongest epidemiologic evidence relating dietary factors to the risk of coronary heart disease has been provided by three large prospective studies: the Health Professionals Follow-up Study, the Alpha-Tocopherol Beta-Carotene Cancer Prevention Study, and the Nurses' Health Study. Those studies assessed the intake of trans fatty acids using detailed food-frequency questionnaires whose results were validated by comparison with the composition of adipose tissue or food diaries. Each of these studies reported an adverse effect of trans fatty acids. The relative risk of coronary heart disease associated with an absolute increase of 2 percent in the intake of trans fatty acids was 1.36 (95 percent confidence interval, 1.03 to 1.81) in the Health Professionals Follow-up Study, 1.14 (95 percent confidence interval, 0.96 to 1.35) in the Alpha-Tocopherol Beta-Carotene Cancer Prevention Study, and 1.93 (95 percent confidence interval, 1.43 to 2.61) in the Nurses' Health Study. The higher relative risk in the Nurses' Health Study may have resulted from the fact that there were four dietary measurements during the follow-up period, thereby reducing the degree of error in assessing trans-fatty-acid consumption. In these three cohorts, the relative risks were higher than those for saturated-fat consumption. For example, in the Nurses' Health Study, replacing 5 percent of energy intake from saturated fat with unsaturated fat was associated with a 42 percent decrease in the risk of coronary heart disease, whereas replacing 2 percent of energy intake from trans fatty acids with cis fatty acids was associated with a 53 percent decrease in the risk.
These studies have been criticized on the grounds that measurements of the intake of trans fatty acids were unreliable however, random errors in measuring the intake would only have led to an underestimation of the association with the risk of coronary heart disease. It has also been suggested that the observed associations resulted from a shift from the use of butter to the use of margarine among high-risk subjects. If so, the association between the intake of trans fatty acids and the risk of coronary heart disease should have been weaker among subjects with stable margarine consumption and stronger during the first few years of follow-up. However, in the Nurses' Health Study,) the exclusion of women who changed their diet before the beginning of the study strengthened the association. Moreover, consumption of foods high in trans fatty acids such as cookies, which are hardly perceived as healthy, was also positively associated with the risk of coronary heart disease.
Confounding as a result of unmeasured or poorly measured risk factors is a potential problem in any observational study, but these associations were adjusted for many risk factors related to diet and lifestyle, and no credible confounding factor has been identified. Adjustment for the intake of dietary fiber attenuated the relation of trans-fatty-acid intake to the risk of coronary heart disease in the Health Professionals Follow-up Study, but not in the other two studies (and Hu FB: personal communication). Thus, prospective studies provide consistent evidence that the consumption of trans fatty acids increases the risk of coronary heart disease. The observed relative risks of coronary heart disease were larger than one might predict from the effects of trans fatty acids on LDL and HDL cholesterol levels alone. The increases in triglyceride and Lp(a) lipoprotein levels account for only a small increase in risk; therefore, other mechanisms may be involved.
Metabolic and epidemiologic studies indicate an adverse effect of trans fatty acids on the risk of coronary heart disease. Furthermore, on a per-gram basis, the adverse effect of trans fatty acids appears to be stronger than that of saturated fatty acids. When ingredients with no known nutritional benefit are added to foods, a low threshold for evidence of harm should be adopted, and it should be the responsibility of food manufacturers to show that their products are safe. In Europe, producers have responded rapidly by developing margarines free of trans fatty acids that are also low in saturated fats. These products are also becoming available in the United States, but heavily hydrogenated stick margarines still retain a large share of the market. In the United States, only 25 percent to 37 percent of the intake of trans fatty acids from hydrogenated vegetable oil comes from margarines; the remainder comes from baked goods, fried fast foods, and other prepared foods. It is more difficult to replace trans fatty acids with healthier fats in such products than in margarines, but the switch could be encouraged by a change in federal regulations.
Current U.S. regulations provide an incentive to manufacturers to produce foods high in trans fatty acids because food labels are not required to include the amount of trans fatty acids. Many scientists agree that the amount of trans fatty acids should be stated on food labels. One simple option is to combine this information with the saturated-fat content. This approach, however, ignores the observation that the intake of trans fatty acids is associated with a higher risk of coronary heart disease than is the intake of saturated fatty acids.
Although changes in labeling are important, they are not enough. Many fast foods contain high levels of trans fatty acids, are exempt from labeling regulations, and can even be advertised as cholesterol-free and cooked in vegetable oil. For example, the consumption of one doughnut at breakfast (3.2 g of trans fatty acids) and a large order of french fries at lunch (6.8 g of trans fatty acids) adds 10 g of trans fatty acids to one's diet and represents 5 percent of the total energy intake on an 1800-calorie diet -- and neither product needs to be labeled.
Five years ago, it became evident that consumption of trans fatty acids adversely affects blood lipid levels. Subsequent studies have confirmed these metabolic findings and strengthened epidemiologic evidence of an important increase in the risk of coronary heart disease with the consumption of trans fatty acids. These data highlight the need for labeling requirements that include fast foods. Given the proper incentives, the food industry could replace a large proportion of the partially hydrogenated fats used in foods and food preparation with unhydrogenated oils. Such a change would substantially reduce the risk of coronary heart disease at a moderate cost, without requiring major efforts focused on education and behavioral modification.
Alberto Ascherio, M.D., Dr.P.H.
Harvard School of Public Health
Boston, MA 02115
Martijn B. Katan, Ph.D.
Peter L. Zock, Ph.D.
Wageningen Center for Food Sciences
6703 HD Wageningen, the Netherlands
Meir J. Stampfer, M.D., Dr.P.H.
Walter C. Willett, M.D., Dr.P.H.
Harvard School of Public Health
Boston, MA 02115
Address reprint requests to Dr. Ascherio at the Department of Nutrition, Harvard School of Public Health, 665 Huntington Ave., Boston, MA 02115, or at alberto.ascherio@channing.harvard.edu.
We are indebted to Jill Arnold for her expert assistance with the manuscript.