Early trials that assessed various lipid-lowering therapies reported little or no angiographic evidence of efficacy, in large part because of modest changes in serum lipids and lipoprotein levels, small sample sizes, and other methodologic problems [2-4]. However, beginning with the Cholesterol Lowering Atherosclerosis Study (CLAS), seven lipid-lowering studies with interventions ranging from multifactorial lifestyle modification, to diet and exercise or diet and either monotherapy or combination drug therapy, to ileal bypass surgery, have shown clear reductions in the progression of atherosclerotic disease or actual lesion regression [5-12], or both. In three trials, clinical coronary events were significantly reduced [7, 8, 10]. The Program on the Surgical Control of the Hyperlipidemias (POSCH), which showed that cholesterol lowering (through partial iliac bypass surgery) had a beneficial effect on coronary artery lesions [8], recently found that the 3-year global change score, an overall consensus judgment of angiographic changes in coronary lesions determined by blinded panels of experts, was predictive of subsequent coronary events (P < 0.0001), fatal coronary events (P = 0.003), and all-cause mortality (P = 0.01) over 3 to 10 years [13]. Although not in the context of a lipid-lowering trial, Waters and colleagues [14] found that patients with angiographic progression of disease at 2 years (by quantitative coronary angiography) also had a significantly increased risk for clinical coronary events during a 44-month follow-up period. In CLAS, we showed that diet in conjunction with colestipol and niacin therapy had a beneficial effect on coronary artery lesions at 2 and 4 years based on the global change score [5, 15] and at 2 years based on quantitative coronary angiography [16]; CLAS was the only angiographic trial to have used both the findings of quantitative coronary angiography and the global change score as end points.

We present the results of a second angiographic trial using these two end points. The Monitored Atherosclerosis Regression Study (MARS) was a double-blind, placebo-controlled, randomized trial that tested the reduction of low-density lipoprotein (LDL) cholesterol levels using lovastatin, an inhibitor of 3-hydroxy-3-methylglutaryl coenzyme A (HMG CoA) reductase. We compare our results regarding the effect of reducing lipid levels on coronary artery status with those of the CLAS and POSCH trials.

Methods

Top Methods Results Discussion Author & Article Info References

Patients were enrolled in the study if they were younger than 70 years and had coronary artery disease in at least two segments, with at least one segment showing diameter stenosis of 50% or more (but not total occlusion) and unaltered by percutaneous transluminal coronary angioplasty. Total cholesterol levels ranged from 4.92 to 7.64 mmol/L (190 to 295 mg/dL). Major exclusion criteria included hypertension (diastolic blood pressure >115 mm Hg or, if the patient was receiving treatment, >100 mm Hg), diabetes mellitus, and the use of lipid-lowering drugs within 2 months of randomization. Women were excluded if they were premenopausal, unless they had undergone surgical sterilization. Candidates for coronary artery bypass graft surgery were excluded, but candidates for percutaneous transluminal coronary angioplasty were not.

Between 1985 and 1989, 270 patients were randomly assigned within one of eight blocks, defined by sex, smoking status (current smoker or nonsmoker), and plasma total cholesterol ( 6.22 mmol/L [240 mg/dL] or >6.22 mmol/L) to receive either lovastatin, 40 mg twice a day, or placebo. Treatment groups had identical targets for cholesterol and fat intake (daily cholesterol intake 250 mg; 27% of calories as fat, with saturated fat constituting 7% of total fat calories and monounsaturated and polyunsaturated fats each accounting for 10% of fat calories). In 58% of cases, patients were recruited through cardiac catheterization laboratories and had undergone baseline coronary arteriography for clinical indications. The percutaneous femoral technique was used, and sufficient right and left anterior oblique views were obtained to show all lesions clearly [11]. During the trial, back-titration of the lovastatin dosage, which was necessary if two consecutive total cholesterol levels were less than 3.11 mmol/L (120 mg/dL) or if one total cholesterol level was less than 2.85 mmol/L (110 mg/dL), was done once for 15 patients and twice for 3 patients. To maintain the blind, the dosage for a matched placebo recipient was also back-titrated.

Lipid, lipoprotein, and apolipoprotein levels were measured using standardized laboratory procedures [11]; safety laboratory tests were done; concomitant medications, drug and diet compliance [11], lenticular opacification [17], and symptoms and adverse events were assessed during the trial. A follow-up angiogram (n = 247), done 2 years after randomization, duplicated the protocol followed for the baseline angiogram, including nitroglycerin use, except in the 25 patients (8 receiving lovastatin and 17 receiving placebo) for whom nitroglycerin was required only at follow-up because of angina. Lesions in patients with a nitroglycerin mismatch were excluded from the quantitative coronary analyses. Baseline films and procedure reports, reviewed before follow-up angiography was done, ensured matching of all variables including the sequence of arteriographic projections, roentgenographic field, and catheter size. An optional double-blind 2-year extension of the original randomized therapy (MARS-II) was accepted by 70% of the patients (98 receiving lovastatin and 75 receiving placebo). Results of MARS-II have not yet been reported.

Evaluation of Coronary Angiograms

All evaluable film pairs (n = 246) showing identical coronary artery views, with treatment allocation and temporal order masked, were evaluated by an expert panel of two angiographers and a moderator [11, 18]. (One placebo recipient did not have an evaluable angiogram by panel procedures.) Using the first film, the panel reached a consensus on lesion identification and percent diameter stenosis, and these data were recorded by the moderator. Using the second film, the panel reached a consensus on the degree of change in each lesion and obtained a global change score (0 [no demonstrable change] to 3 [extreme change]) that integrated panel-based visual changes. A direction for change (- for regression, + for progression) was subsequently assigned by the study statistician when the temporal blind was broken.

Quantitative coronary angiography analyses were done by a single technician blinded to treatment but not to temporal ordering [11, 19]. Film pairs (n = 220) were processed in tandem using dual projectors to match frames for orientation and degree of contrast filling, and arterial segments were defined from branch to branch. Three sequential frames exposed during end-diastole were digitized when possible; if such a sequence was unavailable, three sequential frames from other phases of the cardiac cycle were digitized. The right anterior oblique view was preferred for the quantitative coronary angiography analysis, but other views were substituted if they were superior. Percent diameter stenosis and minimum lumen diameter were measured in each lesion identified by the panel and in lesions identified by the quantitative coronary angiography analyst but not by the panel; these latter lesions tended to be in smaller segments and were less severe. Edge coordinates were corrected for pin-cushion distortion (artifactual coronary artery edge distortion seen in angiographic films) measured from the image of a 1-cm anteroposterior grid filmed at the beginning of each angiogram. Each end point was averaged over three sequential frames.

The primary end point was the average (per-patient) change from baseline in percent diameter stenosis in all lesions that showed 20% stenosis at baseline or at follow-up as evaluated by quantitative coronary angiography. Power calculations based on this end point (and standard deviations estimated to be in the range of 6% to 8%) indicated that an effective sample size of 250 patients had at least an 80% power to detect a treatment difference of 2% to 3% in diameter stenosis at the 0.05 significance level (two-sided) [11].

Three equally weighted secondary end points were average (per-patient) change in minimum lumen diameter (assessed by quantitative coronary angiography), the global change score (assessed by the panel), and the proportion of patients with progression or regression of disease (assessed by quantitative coronary angiography). A patient defined as having progression or regression had lesion change of one type only, either progressing or regressing, but not both types; a patient with a mixed lesion response (that is, with both progressing and regressing lesions) was considered unchanged. A progressing or regressing lesion was defined by a change in percent diameter stenosis of 12% or greater. This cutoff is equal to twice the standard deviation (5.7%) elicited by analysis of repeated angiograms (at the beginning and end of the same angiographic session) of the same lesion in 17 patients (55 lesions) in the MARS study. New lesions and new total occlusions were not counted as progressing lesions; recanalizations were not counted as regressing lesions. Analyses of change in both percent diameter stenosis and minimum lumen diameter were specified a priori for large ( 50% at baseline) lesions [11].

For percent diameter stenosis determined by either quantitative coronary angiography or the panel, tertiary end points included the proportion of patients with 1 or more new lesions and 1 or more new total occlusions. For lesions identified by the panel, progression or regression was based on the consensus of the panel readers using the Delphi procedure described above.

Statistical Analysis

All analyses used the all-patients-treated approach (that is, patients who had an early follow-up angiogram were included in the analysis; patients who did not have a follow-up angiogram were excluded). The Student t-test, the Wilcoxon rank-sum test, the Fisher exact test, and the chi-square test were used to compare treatment groups at baseline; to evaluate within- and between-group changes in lipid and apolipoprotein levels; to compare treatment groups regarding global change score, the proportions of progressing and regressing patients, and the proportions of patients with at least one new lesion and at least one new total occlusion; and to compare treatment groups regarding the incidence of adverse experiences.

The average per-patient change in percent diameter stenosis and minimum lumen diameter were evaluated using analysis of covariance. Planned covariates were percent diameter stenosis at baseline, number of lesions at baseline, and clinical center; only percent diameter stenosis at baseline was found to be significantly related to outcome. Analyses were carried out for all lesions, small lesions (<50% stenosis), and large lesions ( 50% stenosis) at baseline.

Results

Top Methods Results Discussion Author & Article Info References

Baseline Characteristics

Of the 270 randomized patients, 247 (91%) were male; 216 (80%) were current or former smokers and 173 (64%) had a total cholesterol level of less than 240 mg/dL (6.20 mmol/L). The average age was 58 years (range, 37 to 67 years). The average percentage ideal body weight was 122%, and the average blood pressure was 125 mm Hg (systolic) and 80 mm Hg (diastolic). One hundred twenty-three patients (46%) gave a history of hypertension; 112 (41%) had angina pectoris, and 163 (60%) had previous myocardial infarction. Thirty-five patients (13%) had previous percutaneous transluminal coronary angioplasty, and 51 (19%) had had bypass surgery. The mean duration of time from performance of the baseline angiogram to randomization was 2.1 months (range, 0.4 to 5.4 months). No statistical differences between lovastatin and placebo recipients were found regarding demographic or cardiovascular risk factors.

The mean duration from randomization to performance of the second angiogram was 2.2 years (range, 0.6 to 2.7 years) and did not differ between the treatment groups. Nine patients (6 receiving lovastatin, 3 receiving placebo) did not have a second angiogram for medical reasons, and 14 patients (5 receiving lovastatin, 9 receiving placebo) did not have a second angiogram for personal reasons. These patients did not differ statistically from the patients who did have a second angiogram in demographic or cardiovascular risk factors or in lipid and lipoprotein responses to treatment (up to the time of their withdrawal). Further, no statistical differences were found between treatment groups in demographic or cardiovascular risk factors for the 247 patients (123 lovastatin recipients, 124 placebo recipients) who had a repeat angiogram.

A total of 1889 and 3072 lesion pairs in native arteries were assessed by quantitative coronary angiography and the panel, respectively. The average number of lesions per patient was 8.5 based on quantitative coronary angiography and 12.4 based on panel consensus; the averages were 7.2 and 8.6, respectively, for small lesions (<50% stenosis) and 1.5 and 3.8, respectively, for large lesions ( 50% stenosis) (P > 0.20 for all comparisons). The average baseline percent diameter stenosis based on quantitative coronary angiography and panel consensus is shown in Table 1. No statistical differences were found between treatment groups.

Effect of Treatment on Plasma Lipid, Lipoprotein, and Apolipoprotein Levels

The mean baseline and on-trial lipid, lipoprotein, and apolipoprotein levels, as well as percentage change from baseline, in patients who had a follow-up angiogram are shown in Table 2. No statistical differences were found between the groups regarding any of these variables at baseline. For the lovastatin group, total cholesterol, LDL cholesterol, triglyceride and apolipoprotein B levels decreased, and the HDL cholesterol level increased (P < 0.001 for all comparisons); for the placebo group, the total cholesterol level decreased (P < 0.01) and the apolipoprotein B level increased (P < 0.001). Between-group differences were found in total cholesterol, LDL cholesterol, HDL cholesterol, triglyceride, and apolipoprotein B levels (P < 0.001 for all comparisons).

The average prescribed dose was 73 mg/d for lovastatin recipients and 70 mg/d for placebo recipients. Compliance with study medication was not statistically different between the groups (96% in the lovastatin group and 95% in the placebo group); the average actual dose (% compliance x prescribed dose) was 70 mg/d (lovastatin) and 67 mg/d (placebo). No statistical differences were found in the percentages of patients using the following concomitant medications: ß-blocking agents (lovastatin group, 63%; placebo group, 57%); calcium-channel antagonists (lovastatin group, 77%; placebo group, 83%); angiotensin-converting enzyme inhibitors (lovastatin group, 8%; placebo group, 10%); diuretics (lovastatin group, 12%; placebo group, 20%); and aspirin (lovastatin group, 90%; placebo group, 88%).

End Point Analyses

Primary End Point

The mean (±SD) per-patient percent diameter stenosis (adjusted for baseline percent diameter stenosis) based on quantitative coronary angiography Figure 1 increased in both the lovastatin and placebo groups (1.6% ± 6.7% compared with 2.2% ± 6.8%, respectively [P > 0.20]). No statistical differences between treatment groups were found for lesions with stenosis of less than 50% at baseline; however, for lesions with stenosis of 50% or greater at baseline, a mean decrease of 4.1% ± 11.0% occurred with lovastatin therapy compared with a mean increase of 0.9% ± 11.0 with placebo (P = 0.005).

View larger version (18K): [in this window] [in a new window]   Figure 1. Average change in percent diameter stenosis as determined by quantitative coronary angiography. After adjusting for the percent diameter stenosis at baseline, analysis of covariance was carried out for all lesions (114 patients in the lovastatin group, 106 inthe placebo group), small lesions (<50% stenosis) at baseline (112 patients in the lovastatin group, 105 in the placebo group), and large lesions ( 50% stenosis) at baseline (77 patients in the lovastatin group, 79 in the placebo group).

Secondary End Points

The mean (±SD) minimum lumen diameter for all lesions per patient decreased (worsened) 0.03 mm ± 0.21 mm in the lovastatin group and 0.06 mm ± 0.21 mm in the placebo group (P > 0.20). In lesions with stenosis of 50% or more at baseline, lovastatin therapy resulted in an increase (improvement) in mean minimum lumen diameter of 0.13 mm ± 0.35 mm, whereas placebo administration resulted in a decrease of 0.04 mm ± 0.36 mm (P = 0.002).

The mean (±SD) global change score in patients treated with lovastatin was significantly lower (that is, they showed less progression) than in placebo recipients (0.41 ± 1.14 compared with 0.88 ± 1.12, P = 0.002). Differences in the global change score remained evident in the subgroup of patients with a baseline total cholesterol level of 6.22 mmol/L (240 mg/dL) or more (0.46 ± 1.19 [41 lovastatin recipients] compared with 1.11 ± 1.15 [47 placebo recipients], P = 0.01) and in the subgroup with a baseline total cholesterol level less than 6.22 mmol/L (0.39 ± 1.13 [82 lovastatin recipients] compared with 0.74 ± 1.09 [76 placebo recipients], P = 0.05).

Using the 12% cutoff in change in percent diameter stenosis for the quantitative coronary angiography analysis of lesion change, we found fewer patients with progression (29% compared with 41%, P = 0.07) and more patients with regression (23% compared with 12%, P = 0.04) among patients receiving lovastatin than among placebo recipients (Table 1). By the panel-based analysis, 58 patients (47%) receiving lovastatin progressed (global change score > 0) compared with 80 placebo recipients (65%) (P = 0.007). In addition, twice as many patients receiving lovastatin (28 [23%]) as placebo recipients (13 [11%]) had a negative global change score (P < 0.02), indicating regression of disease.

Tertiary End Points

As shown in Table 1, proportionately fewer patients receiving lovastatin than placebo recipients had one or more lesions progressing to total occlusion (P < 0.05 for lesions both identified by quantitative angiography and by the panel). Patients with new occlusions showed no treatment group differences in the baseline number of lesions, overall percent diameter stenosis, and baseline percent diameter stenosis of the lesions that progressed to total occlusion.

Clinical Coronary Events and Clinical and Laboratory Adverse Experiences

Twenty-two lovastatin and 31 placebo recipients had one or more of the following clinical coronary events: myocardial infarction, percutaneous transluminal coronary angioplasty, coronary artery bypass surgery, coronary death, and hospitalization for unstable angina. No statistical differences were observed between groups in the proportions of patients with clinical coronary events.

No statistical differences were observed between treatment groups in the proportions of patients with the following events: death (2 lovastatin recipients and 1 placebo recipient); cancer (6 lovastatin recipients and 5 placebo recipients); withdrawal from the study because of nonfatal adverse events (3 lovastatin recipients and 6 placebo recipients); loss of visual acuity (at least a two-line decrease by Snellen chart examination) (7 lovastatin recipients and 6 placebo recipients); myopathy (none in either group); drug-related elevations (at least three times the upper limit of normal) of aspartate transaminase or alanine transaminase (3 lovastatin recipients and 2 placebo recipients) or persistently elevated aspartate transaminase or alanine transaminase levels (no lovastatin recipients and 2 placebo recipients).

Nonfatal, noncoronary adverse events leading to a patient's withdrawal from the trial included myalgia and anxiety (2 lovastatin recipients) and cerebrovascular accident, hyperthyroidism, hypercholesterolemia, and hypertriglyceridemia (4 placebo recipients). No difference was observed between groups in new onset or worsening of lens opacities according to the standardized Lens Opacities Classification System [17].

Discussion

Top Methods Results Discussion Author & Article Info References

The MARS study is the first angiographic trial to test the effects of monotherapy with an HMG-CoA reductase inhibitor on coronary artery lesions and the second to evaluate angiograms using both quantitative coronary angiography and a panel-reading method. Although the primary end point (change in percent diameter stenosis) was not statistically different between groups, treatment with high-dose lovastatin and a cholesterol-lowering diet slowed the rate of progression of more severe coronary artery lesions (see Figure 1, slowed the overall rate of progression [by global change score], and increased the overall rate of regression (by global change score) (see Table 1). Furthermore, minimum lumen diameter, an absolute measure of lumen width independent of the assumption that adjacent segment diameters are normal, paralleled the results for percent diameter stenosis, and thus, gave an internal consistency to our findings. An external Data and Safety Monitoring Committee [independent of the study sponsors or investigators] recommended discontinuation of the optional 2-year extension of MARS, primarily on the basis of the observed treatment benefit in the global change score end point and on the quantitative coronary angiography end point for lesions with stenosis of 50% or more.

According to both panel reading and quantitative coronary angiography, the proportion of patients with one or more new total occlusions was reduced two- to threefold in the lovastatin group (see Table 1), suggesting the stabilization of severe plaques. Although evidence for a greater treatment effect in more advanced lesions compared with new or early lesions has previously been seen in five cholesterol-lowering angiographic trials [2, 3, 7, 9, 10], the mechanisms underlying these findings remain unknown. Nevertheless, data from these trials are consistent with the observation that the progression rate for untreated, larger lesions is greater than that for milder lesions [20]. Prevention of new closures suggests that depletion of the lipid-rich core by lovastatin resulted in plaques that were less apt to fissure, rupture, and develop thrombotic occlusion [21, 22]. However, evaluation of angiograms alone cannot fully address this question. Other changes related to atherogenesis, including improvements in endothelial cell function and vascular reactivity, have also been proposed as benefits of aggressive lipid-lowering therapy [23, 24].

At the maximal lovastatin dosage used in MARS, reductions in LDL cholesterol and triglyceride levels were similar to those seen in CLAS (40% and 20%, respectively), with lovastatin monotherapy having fewer side effects than therapy with high-dose niacin and colestipol [5]. Although physicians commonly prescribe 20 to 40 mg of lovastatin daily, the findings from MARS indicate that doses of up to 80 mg/d are well-tolerated and effective in lowering LDL cholesterol, which is consistent with previous observations [25].

The MARS study is the third randomized coronary angiographic trial to report treatment efficacy using the global change score as an end point (Table 3). These studies were done in different groups of patients recruited at different times. The evidence from 1104 patients, the largest body of coronary angiographic trial data available with a common outcome measure, shows that diverse but aggressive lipid-lowering therapies (diet and partial ileal bypass surgery [POSCH], diet plus colestipol and niacin (CLAS), and diet plus lovastatin [MARS]) reduce the progression of coronary artery disease. Panel-reading procedures, which were identical in the three trials, have been extensively evaluated using data from CLAS [18]. In addition, in POSCH, the 3-year global change score, a visual assessment of change in coronary lesions, was predictive of combined coronary events (P < 0.0001), fatal coronary events (P = 0.003), and all-cause mortality (P = 0.01) occurring between 3 and 10 years after therapy [13]; we are currently evaluating 10-year ischemic heart disease event rates in CLAS.

As shown in Table 3, reductions of 26% to 32% in total cholesterol and of 42% to 45% in LDL cholesterol resulted in significant and similar improvements in the global change score. In these trials, pretreatment levels of total cholesterol (5.95 to 6.52 mmol/L [230 to 252 mg/dL]) and LDL cholesterol (3.90 to 4.62 mmol/L [151 to 179 mg/dL]) were borderline or high risk according to guidelines of the National Cholesterol Education Program [26]. The aggressive reduction of LDL cholesterol in the three trials, nevertheless, is consistent with the recently revised National Cholesterol Education Program treatment goals for patients with established coronary heart disease and LDL cholesterol levels of 3.36 mmol/L (130 mg/dL) or greater, including a reduction in LDL cholesterol to 2.59 mmol/L (100 mg/dL) or lower [26].

The MARS and CLAS trials permit the comparison of the global change score with quantitative coronary angiography, an independent measure of disease severity. The average changes in percent diameter stenosis by quantitative coronary angiography in MARS were –2.3%,-0.7%, and 4.2%, in patients with a global change score of less than 0 (regression), equal to 0 (unchanged), and greater than 0 (progression), respectively (P < 0.001). In CLAS, the corresponding changes were –4.8%,-0.6%, and 3.8% (P = 0.001). When panelists judged overall coronary status as stable or progressing (global change score 0), changes in percent diameter stenosis in the two trials were virtually identical. In contrast, when panelists found overall regression of coronary disease (global change score <0), a significantly greater change in percent diameter stenosis was found in CLAS (P < 0.001).

The between-trial differences in changes in percent diameter stenosis for patients with global change scores less than 0 may be attributable to a larger treatment effect size [16] for lesions less than 50% at baseline in CLAS (effect size, 0.25 [unpublished data]), three times greater than that observed in MARS (effect size, 0.08, [based on data in Figure 1]). Because the panel-reading and quantitative coronary angiography methods were identical in MARS and CLAS, the observed differences in outcome for smaller lesions could be attributable to differences in study populations (for example, CLAS included slightly more hypercholesterolemic, nonsmoking men with previous coronary bypass surgery) or to on-trial differences such as the changes in HDL cholesterol and apolipoprotein A-I levels. In CLAS, treatment with colestipol and niacin increased the HDL cholesterol level by 37% and the apolipoprotein A-I level by 19%; in MARS, treatment with lovastatin increased these levels by 8% and 3%, respectively (see Table 3).

Concerns about the consistency of improvement in coronary lesion status with LDL cholesterol reduction are addressed by the angiographic evidence from 1104 patients, who were evaluated in an identical manner with the global change score in MARS, CLAS, and POSCH. Three disparate therapies all produced equivalent reductions in LDL cholesterol, which were accompanied by improvement in coronary lesions that was visible to expert angiographers and that was associated (in POSCH) with a reduction in the number of coronary events. Ongoing trials with HMG-CoA reductase inhibitors will address the risk/benefit ratio of these agents in the primary and secondary prevention of coronary heart disease morbidity and mortality.

Appendix: MARS Research Group

University of Southern California, Los Angeles, California: David H. Blankenhorn, MD (Principal Investigator); Linda Cashin-Hemphill, MD, Howard N. Hodis, MD, and Dieter M. Kramsch, MD (Co-investigators); Mary Jo Masteller RN, MS, MPH (Clinic Coordinator); Martha Charlson, MS, RD, Suzanne M.R. Funk, Christine Gesselman, Albert H. Jamentz, MD, Chao-ran Liu, MD, Ci-hua Liu, MD, Lila Pirott, and Linny Zurbrugg (Clinic and Laboratory Staff); Stanley P. Azen, PhD, Laurie D. LaBree, MS, Wendy J. Mack, PhD, Olga Morales, and Janice M. Pogoda, MS (Data Coordinating Center); Alex Sevanian, PhD (Laboratory Director); James Martone, MD, Narsing Rao, MD (Ophthalmologists); Donald W. Crawford, MD (deceased), and Laurie D. LaBree, MS (Internal Safety Monitors); Rosie Baca and Jo Darnall (Administration).

University of Wisconsin, Madison Wisconsin: Laurence W.V. DeBoer, MD, and Peter Hanson, MD (Co-investigators); Laura I. Vailas, MS, RD (Clinic Coordinator); Marcia Bosscher, MPH, RD, Karen Kedrowski, MS, RD, Donald Wiebe, PhD, and Jody Wiker, RN (Clinic and Laboratory Staff); Peter R. Slane (Internal Safety Monitor); David L. DeMets, PhD (Biostatistical Consultant).

Kaiser Permanente Medical Center, Los Angeles, California: Peter R. Mahrer, MD (Co-investigator), Betty A. George (Clinic Coordinator).

Jet Propulsion Laboratory, Pasadena, California: Robert H. Selzer, MS, Anne Shircore, MS, and Paul L. Lee, PhD.

Donner Laboratory, Berkeley, California: Ronald Krauss, MD.

Merck Research Laboratories, Rahway, New Jersey: Laurence J. Hirsch, MD, Roger Greguski, BS, and Robert Zupkis, PhD.

Oklahoma Medical Research Foundation, Oklahoma City, Oklahoma: Petar Alaupovic, PhD.