|
|
| |||||||||||||||||||||||||||||||||||||||
Background A diet low in saturated fat and cholesterol is the standard initial treatment for hypercholesterolemia. However, little quantitative information is available about the efficacy of dietary therapy in clinical practice or about the combined effects of diet and drug therapy.
Methods One hundred eleven outpatients with moderate hypercholesterolemia were treated at five lipid clinics with the National Cholesterol Education Program Step 2 diet (which is low in fat and cholesterol) and lovastatin (20 mg once daily), both alone and together. A diet high in fat and cholesterol and a placebo identical in appearance to the lovastatin were used as the respective controls. Each of the 97 patients completing the study (58 men and 39 women) underwent four consecutive nine-week periods of treatment according to a randomized, balanced design: a high-fat diet-placebo period, a low-fat diet-placebo period, a high-fat diet-lovastatin period, and a low-fat diet-lovastatin period.
Results The level of low-density lipoprotein (LDL) cholesterol was a mean of 5 percent (95 percent confidence interval, 3 to 7 percent) lower during the low-fat diet than during the high-fat diet (P<0.001). With lovastatin therapy as compared with placebo, the reduction was 27 percent. Together, the low-fat diet and lovastatin led to a mean reduction of 32 percent in the level of LDL cholesterol. The level of high-density lipoprotein (HDL) cholesterol fell by 6 percent (95 percent confidence interval, 4 to 8 percent) during the low-fat diet (P<0.001) and rose by 4 percent during treatment with lovastatin (P<0.001). The ratio of LDL to HDL cholesterol and the level of total triglycerides were reduced by lovastatin (P<0.001), but not by the low-fat diet.
Conclusions The effects of the low-fat-low-cholesterol diet and lovastatin on lipoprotein levels were independent and additive. However, the reduction in LDL cholesterol produced by the diet was small, and its benefit was possibly offset by the accompanying reduction in the level of HDL cholesterol.
We studied a free-living population of patients with moderate hypercholesterolemia who chose their own food. We attempted to maximize the effectiveness of the diet by means of extensive dietary advice, reinforcement, and monitoring. We chose the NCEP Step 2 diet1 because of its very low level of saturated fat (<7 percent) and cholesterol (<200 mg per day). Lovastatin (20 mg once daily) was an appropriate choice for the lipid-lowering drug since this agent has been extensively studied2,3,4,5 and is widely prescribed in the United States, accounting for 49 percent of the prescriptions for hypolipidemic drugs in 1991; two thirds of these prescriptions specified a dose of 20 mg of lovastatin per day6.
Methods
Patients
Of 186 patients screened for at least their lipid profile, 111 took part in this study. Each gave written informed consent, and the consent forms and study protocol were approved by each of the institutional review boards of the five participating lipid clinics. Forty-four patients were referred or were already known to the staff of the clinics, 22 were friends or relatives of clinic patients, and 39 joined the study because of media announcements or advertisements; information about the reason for participating was missing for 6 patients. The principal criterion for entry was an LDL cholesterol level between 160 and 200 mg per deciliter (4.14 and 5.17 mmol per liter) while the patient was following his or her current diet. Because the study design required following a high-fat, high-cholesterol diet without drug therapy for nine weeks, patients with clinically apparent atherosclerotic disease were excluded. Other major criteria for exclusion were active liver disease; a body weight 30 percent above ideal; poorly controlled hypertension (blood pressure above 160/95 mm Hg); in men, smoking more than 10 cigarettes a day; in premenopausal women, the ability to conceive, unless conception was unlikely, and treatment with oral contraceptives; and plasma triglyceride levels above 300 mg per deciliter (3.39 mmol per liter).
Study Design
The two treatments were the NCEP Step 2 low-fat, low-cholesterol diet and lovastatin (20 mg once daily) with the evening meal. The control interventions were a high-fat, high-cholesterol diet intended to resemble the average U.S. diet7 and a placebo identical in appearance to lovastatin. The study was double-blinded with respect to the administration of lovastatin and placebo. The dietary and drug treatments were given both alone and together, in four combinations: the high-fat diet with placebo, the low-fat diet with placebo, the high-fat diet with lovastatin, and the low-fat diet with lovastatin. Each patient received all four combinations in 1 of 12 sequences, with each combination being administered for nine weeks without intervening washout periods. The randomization schedule was balanced for both period and carryover effects. To familiarize the patients with the procedures for dietary records, a four-week run-in period during which the patients followed their usual prestudy diets preceded the four consecutive test periods. Thus, for each patient the total duration of the study was 40 weeks. The patients visited the clinic at weeks -4, -2, and 0 (run-in period) and weeks 3, 6, and 9 of each test period. The use of any lipid-lowering drugs was stopped six weeks before the start of the study.
Dietary Therapy and Monitoring
The high-fat control diet was designed to provide approximately 40 percent of calories as fat, with 15 percent saturated, 14 percent monounsaturated, and 8 percent polyunsaturated fatty acids, and a cholesterol intake of 350 to 400 mg per day. The diets were individually prescribed on an isocaloric basis. For the control diet, patients were encouraged to consume whole-fat dairy products, eggs, red meats, bacon, organ meats, chicken with skin, cold cuts, mayonnaise, rich desserts, sauces, and high-fat snack foods. For the NCEP Step 2 diet, the consumption of these foods was limited and low-fat substitutes were recommended in accordance with the guidelines of the NCEP Expert Panel1. In this diet, less than 30 percent of the calories are derived from fat, with less than 7 percent saturated, 10 to 15 percent monounsaturated, and less than 10 percent polyunsaturated fatty acids, and the cholesterol intake is less than 200 mg per day1.
Dietary advice was provided by a coordinated group of registered dietitians certified by the University of Minnesota Nutrition Coordinating Center (Minneapolis). Extensive dietary counseling was given at the beginning of each period to encourage adherence to the newly assigned experimental diet, and was repeated at each subsequent visit. The patients were given dietetic scales and measuring cups and spoons and trained in estimating portion sizes with the use of uniform food models. At each visit except that at week -4, four-day diet diaries were completed by all patients and reviewed by the dietitians, who all used the same checklist and measuring aids. These diaries were first analyzed (usually within one week) by Professional Nutrition Systems (Kansas City, Kans.) to produce a composite diet score, the ratio of ingested saturated fat and cholesterol to calories (the RISCC rating)8. The RISCC rating is the cholesterol-saturated fat index of Connor et al.,9 adjusted for energy intake according to the following equation:
((1.01 x saturated fat [in grams]) + (0.05 x cholesterol [in milligrams])) / (kcal/1000)
This dietary analysis was returned to the dietitians before the patient's next visit and used only to help monitor adherence and to assist in counseling. Subsequently, the diet diaries for weeks 6 and 9 were encoded and analyzed by the Nutrition Coordinating Center; the values were averaged to determine for each period the nutrient composition and RISCC scores that were to be used in the statistical analysis of the study.
For the low-fat diet, the target RISCC score was 12 and the highest acceptable score was 15. For the high-fat diet, the target score was 26 and the lowest acceptable score was 22. The study protocol specified that any period lacking an acceptable RISCC score (average of scores for weeks 6 and 9, as analyzed by the Nutrition Coordinating Center) would be excluded from the analysis.
Measurement of Lipids, Lipoproteins, and Apolipoproteins
At each visit a sample of venous blood was drawn after an overnight fast of 12 to 14 hours and collected in EDTA (1 mg per milliliter); it was centrifuged within 4 hours, the plasma was separated, and the plasma sample was shipped at 4 °C to the central laboratory (Medical Research Laboratories, Cincinnati). Cholesterol and triglycerides were assayed enzymatically with a Hitachi 705 automated analyzer10. The laboratory followed the Standardization Program (Part III) of the Centers for Disease Control and Prevention (CDC) and the National Heart, Lung, and Blood Institute11. High-density lipoprotein (HDL) cholesterol was isolated with the modified procedure for precipitation with heparin and 2 mol of manganese chloride per liter12. The level of LDL cholesterol was calculated with the Friedewald equation13 for use during screening and for comparing values for week 6 with those for week 9 of each period.
At week 9 of each period, ultracentrifugation14,15 was carried out to isolate lipoproteins with densities of 1.006 and 1.125 kg per liter -- i.e., the combined fractions LDL plus HDL (density, >1.006) and the HDL3 subfraction, respectively. Calculation of differences then yielded the levels of very-low-density lipoprotein (VLDL) cholesterol, LDL cholesterol, and the HDL2 subfraction.
Also at week 9 of each period, competitive enzyme-linked immunosorbent assays16,17,18,19 were used to measure the apolipoproteins A-I, A-II, B, and E and lipoprotein(a). Internal quality control was evaluated throughout the study with the use of frozen pools at a minimum of two levels of each apolipoprotein. External quality control was assessed with the use of standardized assays for apolipoproteins A-I and B and lipoprotein(a) proposed by the CDC and the International Union of Immunological Societies20,21.
Apolipoprotein E isoforms were characterized (at any point) in the isolates produced by ultracentrifugation yielding lipids with a density below 1.006 kg per liter, by means of one-dimensional isoelectric focusing22. When interpretation was complicated by the presence of sialate isoforms, two-dimensional isoelectric focusing was performed23. If the amount of apolipoprotein E was insufficient for isoelectric focusing, Western blotting was used24.
Monitoring for Adverse Events
At each visit, the pulse rate, sitting blood pressure, and body weight were recorded, and the patients were questioned about adverse events, which were categorized as definitely, probably, possibly, probably not, or definitely not drug-related. Events in the first three categories were termed "drug-attributable." Routine biochemical and hematologic analyses were performed by the central laboratory at the end of each period.
Statistical Analysis
The primary analysis was performed according to the intention-to-treat principle and included data for all periods for which end-of-period (week 9) data were available. In addition, a per-protocol analysis of changes in lipoprotein levels included only data on patients who completed all four test periods, and excluded data on study periods during which dietary goals were not achieved (see Dietary Therapy and Monitoring, above), as specified by the study protocol. For analyses of lipids, lipoproteins, and body weight, the values for week 9 were used to compare the treatments. Comparisons of nutrient intake were performed as described above.
A general linear-models approach was used25,26. An analysis-of-variance model included the investigator, treatment sequence, patient, and treatment as model variables, and the interaction between period and treatment to test for the goodness of fit. To assess the effect of diet and drug, the treatment effect was partitioned into the effects of the diet, drug, and diet-drug interaction. Additivity was tested by examining the diet-drug interaction.
In a post hoc analysis of the per-protocol data base, which had less statistical power than the primary analysis, some factors potentially affecting the responsiveness to diet were entered into the model individually; the homogeneity of the responsiveness to each factor was tested by examining the interaction between the factor and treatment. The factors assessed were sex, obesity (indicated by a body-mass index [the weight in kilograms divided by the square of the height in meters] of more than 27), previous treatment (i.e., whether the patient had previously received professionally prescribed dietary therapy for hypercholesterolemia), and apolipoprotein E phenotype (patients with the E4/E4 and E4/E3 phenotypes were pooled, and those with the rare E2/E2 phenotype were omitted).
Because of the multiple lipoprotein variables and interactions analyzed, the alpha value for interaction was set at 0.01. The relation between treatment and adverse events was determined with the standard chi-square statistic27. All P values are two-tailed.
Results
Of the 111 patients who entered the study, 97 completed all four test periods. Fourteen patients failed to complete the study; five patients did not complete it for personal reasons, three were lost to follow-up, two left the area, two had problems with following the diets, one had an adverse event, and one had an LDL cholesterol level above 200 mg per deciliter. The data on 6 of these 14 patients were not analyzable (i.e., there were no end-of-period data); data for a total of 17 periods were available for the other 8 patients. For the intention-to-treat analysis, the 6 patients with no analyzable data were excluded (leaving 105 patients). For the per-protocol analysis, only the 97 patients completing the study were included. Of these, 39 patients did not comply with the diets during 1 or more periods; thus, 53 of the 388 periods (97 x 4) were excluded from analysis, as specified in advance (see the Methods section). Tablet counts indicated that compliance with taking lovastatin and placebo was 97 to 99 percent during all four interventions. The base-line characteristics of the patients are shown in Table 1.
|
|
Effects on Lipoproteins
The mean lipid, lipoprotein, and apolipoprotein measurements obtained at the end of the four interventions in the intention-to-treat analysis are shown in Table 3, as well as the mean values for the more important variables in the per-protocol analysis. The two methods of analysis produced similar results.
|
|
Because they had no significant interaction with dietary treatment, sex, obesity, previous treatment, and apolipoprotein E phenotype had no detectable effect on the response of any of the variables to diet.
Adverse Events
One patient was withdrawn from the study because of a rash during the low-fat diet-lovastatin period. Twenty-five of the 111 patients reported 83 minor drug-attributable adverse events (as defined in the Methods section) that occurred outside the run-in period, which are summarized in Table 5. There were no significant differences between the interventions in the incidence of adverse events.
|
Effect of Dietary Therapy
According to analysis of standardized, quantitative dietary records, the diets actually consumed by the patients during each study period were close to those specified in the study protocol, with substantial differences between the high-fat and low-fat diets. Nevertheless, the 5 percent average reduction in the mean levels of total and LDL cholesterol produced by the low-fat diet in the presence or absence of lovastatin was much less than the reduction anticipated by the NCEP Expert Panel1. On the basis of metabolic-ward studies, the Expert Panel predicted an average reduction of about 50 mg per deciliter (1.29 mmol per liter) in the level of total cholesterol during the NCEP Step 2 diet; we observed a mean reduction in the total cholesterol level of only 15 mg per deciliter (0.39 mmol per liter). Our results, however, are consistent with those of many previous controlled studies in outpatients with hypercholesterolemia who received less intensive dietary treatment; most of these studies have shown marginal effects on plasma cholesterol levels28. The effectiveness of the NCEP Step 2 diet has not been previously determined in a controlled study of free-living U.S. outpatients with hypercholesterolemia who chose their meals. More pronounced effects of diets low in saturated fat and cholesterol have been reported in several studies of subjects in institutions or metabolic wards,28,29,30,31 presumably because dietary intake can be more readily controlled under those conditions.
The results of nutrient analysis suggested good dietary adherence; they were obtained by standard methods and were consistent. However, inaccuracies in dietary data may be due to the tendency of patients to report what the dietitian will want to hear, and patients may adhere to their diet better during the days they are completing the diary than at other times. For example, patients completing diet diaries often underestimate their true caloric intake32,33. These problems are inherent in any study in free-living subjects choosing their food. It is possible, therefore, that the differences between the compositions of the two diets were overestimated. The diets used in this study were intended to be isocaloric, but patients often spontaneously consume fewer calories and lose weight when placed on a low-fat diet34,35. Although there was a significant mean difference in weight of 1.4 kg, the estimated difference in energy of approximately 43,000 kcal (180 MJ) between the diets over the nine-week study period should have led to a weight difference of about 5 kg36 if no adaptation such as a change in basal metabolic rate or body water occurred.
Our study devoted considerable effort to encouraging dietary compliance: the diets were administered and monitored by a coordinated group of expert dietitians, certified by the University of Minnesota Nutrition Coordinating Center, who had access to a central dietary-monitoring program; the patients were selected because they were expected to comply, and were given dietary measuring equipment and trained in estimating portion sizes from uniform food models. Therefore, if the patients in this study were unable to comply adequately with the diets, the typical patients treated by most practitioners will probably have even more difficulty with compliance. Consequently, the results obtained in this study may overestimate the average effect of prescribing a highly restrictive lipid-lowering diet in routine clinical practice. Nevertheless, individual patients who are highly motivated may achieve results that are much better than average. In a recent study (the St. Thomas' Atherosclerosis Regression Study),37 a small group of patients with coronary disease who were selected on the basis of their responsiveness to cholestyramine had both a mean reduction of 16 percent in the LDL cholesterol level and improvement in the condition of their coronary lesions after they followed a diet similar to the NCEP Step 2 diet for three years.
The reduction in the level of LDL cholesterol produced by the NCEP Step 2 diet was matched by a similar percentage decrease in the level of HDL cholesterol, so that the ratio of LDL to HDL cholesterol did not change significantly. Weight loss temporarily reduces the level of HDL cholesterol, but the estimated mean decrease with a loss of 1.4 kg is only 0.4 mg per deciliter (0.01 mmol per liter)38. A reduction in the level of HDL cholesterol has been noted in some7,29,31,39,40 but not all37 previous studies of diets low in saturated fat and cholesterol. This reduction can probably be attenuated or reversed by exercise41 and, in obese subjects, by sustained weight loss38. Because many studies in defined population groups have shown that the level of HDL cholesterol is correlated inversely with the incidence of coronary disease,42 the net effect of dietary reduction of both LDL and HDL cholesterol levels on the risk of cardiovascular disease is unclear43. Metabolic-ward studies7,30,40 suggest that replacing saturated fat with monounsaturated fat, instead of replacing it with carbohydrates, lowers LDL cholesterol levels effectively while producing little change in HDL cholesterol levels. A diet of this type is palatable and traditional in Mediterranean countries, where coronary disease is relatively infrequent44. Further work is needed to improve the understanding of effects of diet on HDL cholesterol, to determine how best to maintain HDL cholesterol levels on an outpatient basis, and to optimize dietary therapy by measuring changes in coronary lesions37.
The Combined Effects of Diet and Drug Therapy
Our results show clearly that the effects of diet and lovastatin are independent of each other and additive. Cobb et al. reached a similar conclusion in their study of subjects given lovastatin under metabolic-ward conditions,31 but Clifton et al. have recently reported in a small study that dietary therapy was less effective in outpatients when the closely related drug simvastatin was administered39. The additional reduction in LDL cholesterol levels produced by a strict diet and lovastatin in a dose of 20 mg once a day appears equivalent to the reduction produced by doubling this dose2. On the other hand, the diet undercuts the small drug-induced increase in HDL cholesterol levels.
We conclude that the effects of a strict diet and lovastatin were independent and additive. However, the reduction in LDL cholesterol levels produced by diet was small, and its benefit was possibly offset by the accompanying reduction in HDL cholesterol levels.
Supported by a grant from Merck Research Laboratories.
We are indebted to the patients who participated in this study for their dedication to the demands of the protocol, and to Dr. M. Buzzard, J.R. Nelson, and M. Stevens (Nutrition Coordinating Center, Minneapolis) and L. Votaw and G. Peterson (Professional Nutrition Systems, Kansas City, Kans.) for their analysis of the dietary records.
Source Information
From the Heart Disease Prevention Clinic, Minneapolis (D.B.H., E.Q.); the Christ Hospital Cardiovascular Research Center, Cincinnati (E.A.S., S.H.); the University of Kansas Medical Center, Kansas City (C.A.D., W.S.H.); the Medical College of Georgia, Augusta (E.B.F., S.B.L.); George Washington University, Washington, D.
(V.T.M., B.K.B.); Merck Research Laboratories, Rahway, N.J. (J.A.T., A.M.T.); Medical Research Laboratories, Cincinnati (P.M.L.); and Professional Nutrition Systems, Kansas City, Kans. (J.H.). The following persons also participated in the study: Heart Disease Prevention Clinic, Minneapolis -- J. Peters and K. Gardner; Christ Hospital Cardiovascular Research Center, Cincinnati -- D.M. Black, G.E. Lamkin, and S. Ames; University of Kansas Medical Center, Kansas City -- P. Krehbiel and S. Horniman; Medical College of Georgia, Augusta -- T.T. Kuske and J.M. Greene; George Washington University, Washington, D. -- D.B. Stoy and G. Gasparis; and Medical Research Laboratories, Cincinnati -- P.M. Steiner.
Address reprint requests to Dr. Hunninghake at the Heart Disease Prevention Clinic, 151 Variety Club Heart & Research Center, 401 E. River Rd., Minneapolis, MN 55455.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Treatment of and Screening for Hyperlipidemia
Ornish D., Brown S. E., Kottke B. A., Shea S., Barth J. D., Bryan G. K., Hokanson J. E., Austin M. A., Ginsberg H. N., Tall A. R., Deckelbaum R. J., Hunninghake D. B., Criqui M. H., Heiss G., Sox H. C.
Extract |
Full Text
N Engl J Med 1993;
329:1124-1128, Oct 7, 1993.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
