The effect of fish oil supplements on atherogenesis is controversial, especially when fish oil does not lower plasma cholesterol. Some studies in swine have shown that a fish oil supplement to a butter-cholesterol diet reduces atherogenesis. The fish oil supplement also frequently reduces average plasma cholesterol levels. The reduction in lesion size has been shown to be greater than can be expected from average plasma cholesterol reductions, if a linear relationship between lesion size and plasma cholesterol was assumed. However, in an experiment in which we equalized time-weighted average plasma cholesterol levels, there was no significant reduction in lesion size in the fish oil supplemented group. This led us to question the validity of the linear relationship between lesion size and plasma cholesterol level. In this study we have combined the results of eight study blocks with a total of 76 swine fed a similar hyperlipidemic, butter-cholesterol diet. Of these, 24 received a fish oil supplement (BT + fish oil) and 52 swine received no fish oil supplement (BT). The average lesion size as measured by nuclear profiles per cross section of a fixed site in the abdominal aorta (ABNpCx) was 7704 ± 778 (mean ± SEM) for the BT group and 2360 ± 1145 for the BT + fish oil group. Total plasma cholesterol levels were measured at the outset and at monthly intervals until sacrifice. For each animal we obtained a time-weighted average based on the trapezoidal rule. The average of all animals of the time-weighted average plasma cholesterol level was 630 ± 19 mg/dl for the BT group and 486 ± 27 mg/dl for the BT + fish oil group. Using the so-called Box-Cox method for transformation on the combined data set, we show that an exponential relationship between ABNpCx and time-weighted average plasma cholesterol level is a highly appropriate model and significant (P < 0.0001). Neither the fish oil nor the block effect on lesion size is significant above and beyond the differences that might be due to plasma cholesterol levels.

We have investigated in swine the effect of fish oil additives to a butter-cholesterol hyperlipidemic diet (BT) on atherogenesis and thrombogenesis when average plasma cholesterol levels were kept similar in fish oil-treated and untreated BT groups. The studies included evaluation of lesion sizes and cell numbers, counts of adherent monocytes over lesions, and counts of platelet clumps (microthrombi) over lesions either attached directly to endothelium or to adherent monocytes. Anatomic sites studied for lesion development were the left anterior descending coronary artery (LAD), the distal 15 of the abdominal aorta, and a proximal portion of the thoracic aorta. Counts of attached monocytes and platelet clumps were made by scanning electron microscopy only for the LAD and expressed per mm2 of surface. The most striking new result was in regard to the platelet clumps. These were reduced by the fish oil from 996 ± 295/mm2 in the untreated BT group to 313 ± 59 and 364 ± 105 in BT + cod liver oil and BT + menhaden oil groups, respectively. Most of the platelet clumps were adherent to attached monocytes in all groups and the number of attached monocytes were greatly reduced by the fish oil additive. Thus there were close relationships among platelet clumps, monocytes, and lesion endothelium. Numbers of attachments over nonlesion endothelium were much less than those over lesions in all dietary groups. The most surprising result was the lack of retardation of lesion growth by the fish oil additives in spite of the reduction in attached monocytes and platelet clumps. In previous studies where the high plasma cholesterol levels in the BT swine had been modestly reduced (about 25%) there had been a marked retardation of lesion growth. The current result suggests that plasma cholesterol is the major factor controlling lesion growth in this model though under milder conditions and longer observation periods other factors might become apparent.

This study shows that frank endothelial denudation, as measured by scanning electron microscopy, is associated with coronary artery atherosclerotic lesions in swine fed hyperlipidemic diets for 9 or 18 months. The largest areas of endothelial cell denudation were found to be associated with the most advanced atherosclerotic lesions, and lesser areas of denudation were present in less severely atherosclerotic arteries. Overall, the study suggests that frank endothelial denudation in swine coronaries occurs secondarily to advanced lesion development, and is proportionate to the amount of lesion present. In contrast attachment of monocytes to the endothelial surface was not proportionate to the amount of lesion present. While the number of attached monocytes after 9 or 18 months of a hyperlipidemic diet was much greater than in the mash-fed swine, there was no difference between the two hyperlipidemic groups. In both hyperlipidemic groups, the attached monocytes were concentrated over lesion areas. By scanning electron microscopy, most attached monocytes in these perfused coronary arteries showed a slender foot process extending into a gap between endothelial cells. Endothelial cell turnover, as measured by tritiated thymidine labelling indices, increased with the amount of coronary artery atherosclerosis. However, it did not become significantly greater than in mash-fed swine until advanced lesions developed.

We have investigated several aspects of endothelial cell (EC) behavior during the initiation and early development of intimal cell mass (ICM)-derived atherosclerotic lesions in the distal abdominal aortas of young swine fed hyperlipidemic (HL) diets for 0, 14, 49, or 90 days. By scanning electron microscopy no breaks in endothelial integrity or other abnormalities were observed even at 90 days on diet after lesions were well established. Also, counts of leukocytes adherent to the endothelium by both scanning and light microscopy revealed no greater numbers in HL than in mash control swine. Estimates of individual EC losses over ICM-lesions in the HL swine (based on calculations from tritiated thymidine labeling indices and EC growth rates determined by counts) suggested a loss of approximately one per 100 EC/day. This loss over ICM lesion was not significantly greater than that over ICM in mash controls, but was significantly greater than that of EC not over ICM lesions in the same HL animal. In any event, the estimated loss seems to be too trivial for the endothelial barrier to be compromised even transiently in a biologically significant fashion. A significant correlation was observed between tritiated thymidine labeling indices of cells within the ICM lesions and of those of the overlying EC that was not observed with the ICM in the controls. Possibly the positive correlation may be a result of the abluminal surface of the overlying EC being exposed to the abnormal milieu of the ICM lesion. It is emphasized that results reported here apply only to ICM-derived lesions at an early stage of development and that they do not contradict results obtained by others with other lesion types such as those derived from monocytes. Furthermore, in later stages of development of ICM-derived lesions in the same model we know that extensive endothelial cell damage can be demonstrated. Also, functional changes in endothelial permeability may have been present in early stages that would not have been detected with the methods used in this study.

In the swine abdominal aorta, most atherosclerotic lesions arise from naturally occuring collections of intimal cells called intimal cell masses. The main objective of this study was to determine if the tritiated thymidine labeling index of endothelial cells lying over intimal cell masses was greater than that of endothelial cells not over intimal cell masses. A higher endothelial cell labeling index over intimal cell masses would indicate possibly a greater turnover in the area. Such a finding would suggest possible transient but repeated breaks in the endothelial cell barrier that might contribute to the initiation of the atherosclerotic lesion in intimal cell masses.

In an earlier study of swine fed hyperlipidemic (HL) diets containing either butter or corn oil we found wide differences in atherogenic intimal proliferative responses among the swine, especially in the corn oil group, that could not be entirely accounted for by differences in serum cholesterol levels. We hypothesized that there might be differences in some lipoprotein constituents other than total cholesterol that might better account for the differences in intimal proliferative responses. In the current experiment, 11 swine were fed HL diets containing butter (n = 5) or corn oil (n = 6). At 90 days on HL diet, plasma was obtained for lipoprotein fractionation. At 120 days the swine were killed and the lesion areas and number of nuclear profiles in the intima or lesions in multiple cross-sections of the distal abdominal aorta were determined under light microscopy. Tritiated thymidine labeling indices of the intima or lesion cells were also determined. On comparing the butter and corn oil groups, only IDL-cholesterol and tritiated thymidine labeling indices of the cells in the lesions were significantly different. Based on the number of nuclear profiles per cross-section (Np/Cx) 7 were classified as high intimal proliferative responders (5 butter, 2 corn oil) and 4 as low responders (4 corn oil). The high-proliferative responders had plasma IDL levels 6-fold greater than those of the low responders with no overlaps. LDL, VLDL and HDL were not significantly different between the high- and low-proliferative responders. Among correlation coefficients between Np/Cx and plasma lipoprotein variables for all swine (n = 11), IDL cholesterol level was best correlated with Np/Cx. However, in the high responder group LDL values showed a highly significant correlation with the amount of intimal proliferation.

Hyperlipidemic diet-induced atherosclerotic lesions show a predilection for locating at the site of normal collections of intimal cells in the swine abdominal aorta. This implies that preexisting intimal cell masses (ICM) evolve into early diet-induced lesions. This study characterizes in detail the distal abdominal aortic ICM of young normal swine as a basis for a better understanding of lesion development. Seventeen male Yorkshire swine were used, from neonates to 20 weeks of age. Only three of six neonates showed multilayered ICM, the remaining three showed scattered individual cells or a single layer of cells. All older swine had multilayered ICM at a predictable ventral site corresponding to Evans blue in vivo staining. The majority of cells counted by electron microscopy were smooth muscle (89 to 94%), the remaining were macrophages (1 to 8%) and poorly differentiated cells (3 to 5%). Weanling and older swine showed high proportions of contractile smooth muscle cells. Ergastoplasm-rich smooth muscle cells comprised nearly half of the neonatal population. Stainable lipids were demonstrable in 20-week-olds, but not in younger swine. No foam cells were seen, but cells containing a few droplets totaled 2%; macrophages comprised a high proportion of these. Pyknotic dead cells were rare, being found in the neonate and 20-week-old swine. Mitoses were observed in two ergastoplasm-rich smooth muscle cells in the neonate and one each in contractile smooth muscle cells of a 14- and a 20-week-old swine. Saddlebag-shaped, atypical-appearing cells were present in all age groups, involving all three cell types, and ranging from 1 to 3%. These represented either bridged nuclei and cytoplasm or more likely artefactual distortion of relatively normal cells.

In a previous study the authors demonstrated monotypic foci, all of timidus type, in the atherosclerotic lesions and normal aortic tissue of glucose-6-phosphate dehydrogenase (G-6-PD) mosaic hares fed 25-hydroxycholesterol or triol in addition to a high-cholesterol diet. Other hares fed the high-cholesterol diet without the cholesterol oxidation product additives did not develop monotypic foci. In the current study the question asked was whether ditypic samples from lesions or aortic mediae of the same hares fed the additives would show a higher percentage of timidus type G-6-PD than the hares fed the cholesterol alone. The results showed a significantly higher percentage of timidus type in both lesions and mediae in the groups fed either 25-hydroxycholesterol or triol than in the group fed cholesterol alone. The observations suggest that the change from ditypism to monotypism reported earlier with these cholesterol oxidation products was the result of progressive elimination of the europaeus type to leave only the timidus G-6-PD. This is consistent with the "selective advantage of one phenotype over the other" hypothesis for explaining the observed monotypism as opposed to the "genetic transformation" hypothesis.

In the current study of swine coronary arteries, we have attempted to quantify the coronary intimal cell masses (ICM) with regard to two aspects: (1) percentage of total arterial wall cells (intimal plus medial) that are in the intima, and (2) percentage of arterial wall area occupied by intima. We have studied 4 coronary sites: left main stem (LMS), proximal right (RCA), proximal left circumflex (LCX), and proximal left anterial descending (LAD). In mash-fed ('normal') swine we carried out the study at 3 ages: 2, 5 and 12 months. In addition, we have carried out a similar study in swine fed hyperlipidemic (HL) diets for 3 or approximately 10 months starting at 2 months of age. In the control swine we found approximately 6% of the arterial wall occupied by intima at 2 months of age with no significant increases by 12 months. There were no consistent significant differences among the 4 coronary artery sites with regard to percentage of wall occupied by intima. When the HL diet was fed, there were significant increases in the percentage of the arterial wall occupied by intima (now atherosclerotic lesion) by 3 months on diet and much larger increases by 10 months (up to 87% of wall occupied by intimal lesion) with extensive regions of lipid-rich, calcific, necrotic debris. After 3 months on HL diet the RCA lesions occupied significantly greater percentages of areas than the other 3 sites but this difference had disappeared by 10 months on the HL diet.

The effect of uv irradiation on the recovery of DNA synthesis is examined in a population of hare fibroblasts exhibiting heterozygosity with reference to the X-linked enzyme, glucose-6-phosphate-dehydrogenase (G-6-PD). These cells have been grown from skin explants of a hybrid female cross between Lepus timidus (female) and L. europaeus (male), the former carrying the G-6-PD gene for the slow-moving “T” variant and the latter with the fast-moving “E” variant gene. The hybrid, therefore, exhibits genetic mosaicism due to random inactivation in each cell, of one of the two X chromosomes in the embryonic stage. Exponentially growing cells from 13 fibroblast strains, comprising a wide range of E to T ratios, were exposed to moderately low dose of uv irradiation (6 J/m2). The recovery in DNA synthesis during the 2- to 8-h postirradiation period was calculated as the mean percentage rates of [3H]thymidine incorporated during the time as compared to the unirradiated zero-time controls. The results show a statistically significant positive correlation as determined by linear regression analysis between the levels of E and the rate of recovery in DNA synthesis. This is valid also at the higher dose of uv (21 J/m2). These results strengthen our earlier observations with 25-hydroxycholesterol that in the in vitro system the cell expressing the E variant is perhaps more resistant to cytotoxic agents. This also indicates that various factors contribute to the development of monotypism which include cell growth, cell death, mutation, and selection, to name a few.

The normal aortic tissue of black women heterozygous for glucose-6-phosphate dehydrogenase (G-6-PD) usually consists of two cell phenotypes (mosaicism). By electrophoresis two G-6-PD types are demonstrated (ditypism). Advanced atherosclerotic lesions from such women not infrequently yield samples displaying only one G-6-PD type (monotypism). Possible causes for the monotypism include (1) monoclonal origin of the lesion and (2) selective growth and/or survival advantage of one phenotype over the other. We have been attempting to produce monotypism in the aortas of a hybrid hare model that displays G-6-PD mosaicism in the normal state. In the current study 14 G-6-PD mosaic hares were fed a high-cholesterol diet for 6 to 17 months. All developed extensive moderately severe (as compared to humans) atherosclerotic lesions. No monotypic samples were found. Ten hares were fed (in addition to the standard high-cholesterol diet) small doses of one of two cholesterol oxidation products (25-hydroxycholesterol or triol) for 6 to 21 months. Four of the ten developed monotypic foci in either atherosclerotic lesions or normal aortic tissue or both. The reason for giving the oxidation products was because they are cytotoxic for cells in vitro and one (25-hydroxycholesterol) had been shown to be more toxic for one phenotype than the other with cultured fibroblasts from the mosaic hares (not infrequently resulting in the culture becoming monotypic). We suggest that the monotypism observed in vivo was probably produced by a mechanism similar to that operating in vitro.

The abdominal aortas of five groups of young male Yorkshire swine were studied: (1) 0-day baseline group; (2) hyperlipidemic (HL) group with ballooning; (3) mash group with ballooning; (4) mash group without ballooning; and (5) HL group without ballooning. The last four groups were injected with [3H]thymidine at 270 days and sacrificed subsequently in subsets at intervals up to 75 days in order to study births and deaths (or loss by migration) among cells over the period 270–345 days. However, only in the HL-ballooned group were there enough swine for the isotopic data to be useful for most purposes. In the 0-day baseline group there were 6 ± 2 × 106 cells in intimal cell masses (ICM); in the 270- to 345-day mash group without ballooning the number was 10 ± 2 × 106, which is not a statistically significant increase over 0-day. This supports the hypothesis that in the normal state births and deaths (or loss by migration) among cells in ICM are nearly balanced at least up to 1 year of age. In the 270- to 345-day mash group with ballooning there were 61 ± 12 × 106 cells in the ICM. Thus a single episode of deendothelialization results in tremendous hyperplasia of ICM. However, even the largest ICM (atherosclerotic lesion) in this group showed essentially no necrosis. In the 270- to 345-day HL group with ballooning there were 108 ± 17 × 106 cells in the ICM turned atherosclerotic lesions. In addition an average of one-third of the lesion volume was occupied by lipid-rich, calcific necrotic debris. Thus the HL diet appears to have associated with it both mitogenic and cytotoxic influences on ICM cells. In the 270- to 345-day HL group not ballooned there were 130 ± 30 × 106 lesion cells and lesions were somewhat more extensive and necrotic than in the HL-ballooned group, probably because the former group included by chance more hyperresponders (as regards serum cholesterol values) to the HL diet than the latter. Regardless of this, the data suggest that in this particular model of advanced atherosclerosis the balloon-injury stimulus to proliferation and the HL-diet stimulus are neither additive nor synergistic. Comparisons of data between lesions and media suggest that in the lesions as compared to media (a) cell-loss rates by death and/or migration are higher, (b) cell-increase rates by division are larger, and (c) proportion of cells with shorter cycle times is greater (30% with cell-cycle times <37 days compared with 0% in this category in the media).

In this study fibroblast cultures from the skin of 11 hybrid female hares that were heterozygous for glucose-6-phosphate dehydrogenase (G-6-PD) were grown in a standard culture medium containing successively increasing concentrations of 25-hydroxycholesterol. The hare cross was between Lepus timidus (T) females and Lepus europaeus (E) males. In cultures from 8 of the 11 hares there was a significant shift of T:E ratios toward E with increasing concentrations of 25-hydroxycholesterol. In 3 hares, some cultures with the higher concentrations of additive became phenotypically 100% E. The change of T:E ratios towards E appeared to occur because the 25-hydroxycholesterol was more toxic for the T-type cell. The results in this study contrast with those in a previous study of skin fibroblasts which include these same 11 hares. In the earlier study the fibroblasts were carried through multiple passages in the same standard medium used herein (but with no 25-hydroxycholesterol added). Seven of the eleven eventually exhibited only T-type cells. In 2 there was significant movement toward E but neither showed only E in any passage. The responses of the cells are presumably on a genetic basis. It is interesting but not expected that gene expressions enhancing survival and growth in one cellular environment can do the opposite in another.

This study was designed (1) to quantitate the growth of atherosclerotic lesions in the abdominal aorta of hyperlipidemic ballooned swine over the period 0–165 days postballooning; (2) to investigate birth and death of lesion cells for the period 90–165 days with the aid of isotopic techniques previously used in studying the period 0–97 days postballooning and; (3) to determine some of the effects of changing the hyperlipidemic diet to a low-fat, low-cholesterol diet for the period 60–165 days. Previous results combined with results from this study (for the average of the two dietary groups) regarding (1) indicate an increase of cells in atherosclerotic lesions (and other intimal cell masses) from ∼0.8 × 106 on Day 0 to ∼7.0 × 106 on Day 90 to ∼15.0 × 106 on Day 165. Thus, for Days 0–90 there was a ninefold increase and for Days 90–165 a twofold increase. Expressed differently, the numerical increase for Days 0–90 was ∼6.2 × 106 and that for Days 90–165 was ∼8.0 × 106. Results regarding (2) suggest that cycle times for lesion cells had increased to >75 days from the spectrum beginning as low as 12 days reported previously for the 0- to 90-day period. Also, the lesion cell death rates appear to have dropped from the 50% reported for the 0- to 97-day period to a rate too low to register with the techniques used. Results regarding (3) indicate that giving a mash diet from Day 60 to 165 had no demonstrable effect either on growth of lesions or on divisions and deaths of individual lesion cells. The atherosclerotic lesions in this study almost certainly contained monocytes as well as smooth muscle cells, and the effect that the presence of monocytes would have on our calculations is discussed. The conclusion is that, if monocytes are present in substantial numbers, our calculations of smooth muscle cell divisions are higher than the actual number of divisions and thus the slowdown of division rates of individual lesion smooth muscle cells was even greater than that calculated herein.

Hybrid hares (Lepus europaeus male x Lepus timidus female) were fed intermittently with a cholesterol-containing diet for 3 to 13 months. Their aortas were evaluated for extent of lesions, and for glucose-6-phosphate dehydrogenase (G-6-PD) isozyme patterns in lesions and in normal-appearing aortic media. The red cells of all animals were assayed above 50% for T (the slower-moving form of the enzyme, corresponding to type B in black women, and contributed in the cross by the L. timidus dam). The lesions and normal aortic tissues showed a similar bias toward type T, although values fell slightly below 50% in a few cases. The average percentage of T in lesions and in normal tissues from the same aorta differed significantly in some of the animals, suggesting that cells with one type of active X chromosomes might have some selective advantage over the other in the abnormal environment of the atherosclerotic lesion. The putative selective advantage would have to be related to genes on the X chromosome other than that for G-6-PD, since deviations toward T-type G-6-PD were found in some hares and toward E in others. One hare was fed for 6 months with 25-hydroxycholesterol as well as cholesterol. In this animal (and only in this one) monotypic samples were obtained from lesions, even more frequently from grossly normal regions of the aorta, and from uterus and liver. The monotypes were 100% E in liver, but 100% T in normal aorta, aortic lesions, and uterus. Skin, heart, kidney, spleen, and lung samples were all ditypic. In addition to developing aortic lesions readily on a cholesterol diet, these hybrids show marked stenotic lesions in the extramyocardial coronaries, and are therefore tentatively considered a promising model for the study of atherogenesis.

Skin fibroblast cultures from hybrid hares (Lepus europaeus × Lepus timidus) exhibiting glucose-6-phosphate dehydrogenase (G-6-PD) mosaicism were followed for varying numbers of transfers. The majority of cultures either changed systematically in the ratio of G-6-PD variants, or exhibited, at some transfer, one type only. Changes occurred in both directions: toward higher T:E ratios or toward lower T:E ratios. Some cultures shifted from the apparent monotypic state to the ditypic state (usually from monotype T), suggesting that the apparent monotypes sometimes contain type E cells which fail to be detected, either due to limitations of assay sensitivity or to cytoplasmic factors of varying activity or concentration. The results are somewhat similar to those found in skin fibroblast cultures of heterozygous black female G-6-PD mosaics, and suggest the presence of factors, in one or the other of the X chromosomes, conferring a selective advantage on the cells of the corresponding type.

The temporal variation of HMG-CoA reductase activity in the liver and intestine of swine was investigated. The thin-layer chromatographic method widely used in the assay of the reductase was successfully applied to the porcine enzymes. Parallel circadian rhythms were demonstrated in both hepatic and ileal reductases from mash-fed animals. Peak activity occurred approximately 6 hr after feeding, 2.7-fold over the basal level in the liver, and 1.6-fold in the ileum. A milk-cholesterol diet caused a marked depression of both rhythms (90% in liver, 50% in ileum); however, the hourly variation in activity persisted in both organs. Cholestyramine was found to elevate hepatic activity (2.7-fold throughout the rhythm) without affecting that of the intestine. Clofibrate had no effect on either enzyme at any time during the cycle despite a 34% reduction in serum cholesterol concentrations.

Twenty young (8–10 weeks old and approximately 10 kg) male Yorkshire swine were studied for (1) the effects of the amounts of fat (90, 30, and 0 g butter) added in mash-based diet on serum cholesterol concentrations and fecal neutral steroids and bile acid excretions; (2) the effects of fat in semipurified diets with the same amount of dietary fiber as in mash but derived from wheat bran cereals in the semipurified diets; and (4) the effects of a wheat bran cereal containing a high lignin content. The effects of fat contents on serum cholesterol concentration and fecal steroid excretions were demonstrated in both mash-based diets and in semipurified diets containing wheat bran cereal as the source of dietary fibers. Butter (90 g) added to a mash-based diet resulted in an increase in serum cholesterol concentration and a reduction in fecal steroids as compared to the 0-g butter group. When fat content was varied in a semipurified diet containing wheat bran cereal, serum cholesterol concentrations were elevated during 40 and 20% fat (by calories) diet periods accompanied by the reduction in fecal steroid excretions. On comparing fecal steroid excretions in wheat bran-containing diets and in mash with dietary fiber content of both at 64 g and equal fat contents, wheat bran-containing diets excreted significantly less than mash. Moreover, fecal steroid excretions during wheat bran diets were essentially the same as that of a semipurified diet containing purified cellulose-type fiber as the source of fiber. However, at the low fat content (10% fat by calories), bran-containing diet produced fecal bulk similar to that of mash. A wheat bran cereal containing higher (2.6 times) lignin content than the other did not affect fecal steroid excretion. It appears, therefore, that higher fecal steroid excretion during mash feeding is not principally due to fiber per se but due to factor(s) yet to be named. From our studies on the effects of soy protein, vegetable proteins in mash might be the factor mainly responsible for increased fecal steroid excretions.

The aim of the first part of this study was to find out in swine how much fat could be included in a cholesterol-free semi-purified diet without resulting in changes in serum cholesterol levels or hepatic microsomal HMG-CoA reductase activities from those in swine fed a grain-based mash diet. Graded levels of peanut oil from 5 to 40% by calories were included in a casein-carbohydrate-cellulose based diet and effects compared with those in swine fed mash (10% fat) for 2 weeks. Serum cholesterol levels were essentially unaffected by any of the diets. However, in the semipurified diet-fed swine at all levels of dietary fat hepatic microsomal HMG-CoA reductase activities were reduced to values circa 25% of those in the mash-fed swine. This is as great as the reduction from mash levels of activity produced by high fat, high-cholesterol hyperlipidemic diets.