Predominance of Dense Low-Density Lipoprotein Particles Predicts Angiographic Benefit of Therapy in the Stanford Coronary Risk Intervention Project
Background LDL particles differ in size and density. Individuals with LDL profiles that peak in relatively small, dense particles have been reported to be at increased risk of coronary artery disease. We hypothesized that response to coronary disease therapy in such individuals might differ from response in individuals whose profiles peak in larger, more buoyant LDL. We examined this hypothesis in the Stanford Coronary Risk Intervention Project, an angiographic trial that compared multifactorial risk-reduction intervention with the usual care of physicians.
Methods and Results For 213 men, a bimodal frequency distribution of peak LDL density (g/mL) determined by analytical ultracentrifugation was used to classify baseline LDL profiles as “buoyant mode” (density ≤1.0378) or “dense mode” (density >1.0378). Coronary disease progression after 4 years was assessed by rates of change (mm/y, negative when arteries narrow) of minimum artery diameter. Rates for buoyant-mode subjects were −0.038±0.007 (mean±SEM) in usual care (n=65) and −0.039±0.010 in intervention (n=56; P=.6). Rates for dense-mode subjects were −0.054±0.012 in usual care (n=51) and −0.008±0.009 in intervention (n=41, P=.007). Lipid changes did not account for this difference in angiographic response.
Conclusions Different types of LDL profile may predict different responses to specific therapies, perhaps because metabolic processes determine both LDL profiles and responses to therapies.
Clinical trials using quantitative coronary angiography to measure the effects of therapy on CAD progression have reported a wide variation in these effects and a number of factors related to this variation.1 2 3 4 5 6 7 8 9 10 11 12 These factors include baseline and on-trial measurements as well as changes over the course of the trial of lipid, lipoprotein, and nonlipid variables. Although any of these factors related to angiographic change can contribute to an understanding of the mechanisms of atherosclerosis, those measured at baseline also have potential clinical use in the identification of optimal therapies for individual patients.
LDL comprises distinct subspecies of varying size and density.13 In populations of both healthy and diseased subjects, large variations are seen in the proportions of different subspecies; whereas some LDL profiles show a predominance of small, dense LDL species, others show a predominance of larger, more buoyant LDL. Several studies have reported that lipoprotein profiles marked by relatively high proportions of small, dense LDL are associated with increased risk of CAD and myocardial infarction.14 15 16
LDL profiles with different predominant subspecies suggest the operation of different sets of processes affecting lipoprotein metabolism.17 Thus, we wanted to test the possibility that metabolic intervention may have different effects on CAD progression in subjects with different types of LDL profile. We examined this possibility in SCRIP, 1 2 a 4-year angiographic trial in which intensive multifactorial RR, compared with UC, resulted in reduction of CAD progression, as determined by changes in minimum artery diameter.1
We introduce a classification of LDL particle profiles using a bimodality in the frequency distribution of peak LDL density, as determined by ANUC.18 The modes with lower and higher density values define LDL profiles with a predominance of buoyant or dense LDL, respectively. Because LDL size and density are inversely correlated, we refer to the two types of profiles—“buoyant-mode” and “dense-mode” profiles—as those with a predominance of large, buoyant LDL and small, dense LDL.
Baseline peak LDL densities were used to categorize SCRIP subjects as buoyant-mode or dense-mode subjects. We wanted to determine whether buoyant-mode and dense-mode subjects would respond differently to intervention. Because men and women have different LDL profiles19 20 and the number of female subjects in SCRIP was small (n=29), we analyzed only men; in particular, we analyzed the 213 male subjects with baseline LDL profiles determined by ANUC. On the basis of these analyses, we report that RR was angiographically beneficial for dense-mode subjects but not for buoyant-mode subjects.
We also analyzed angiographic results on the basis of LDL subclass pattern,13 15 a classification of LDL particle size distributions determined by nondenaturing GGE. Patterns A and B are characterized by a predominance of larger or smaller particles, respectively; intermediate patterns are designated pattern I. Although subclass pattern was not as strongly predictive as density mode of benefit of RR, results for pattern A and pattern B subjects were, as expected from the strong correlation of LDL size and density, similar in character to results for buoyant-mode and dense-mode subjects, respectively.
The SCRIP design and results have been described in detail previously.1 Recruitment was through screening at the four hospitals participating in the study for CAD patients who had been referred to those hospitals for coronary angiography, who were <75 years old, who had one or more nonrevascularized major coronary artery segments with lumen narrowing between 5% and 69%, and who had no health problems that would prohibit participation in RR. All subjects signed a consent form and were randomized at baseline to either UC or RR. The study protocol and progress were reviewed before the start and annually by the Stanford University Panel on Human Subjects, by the Committee for the Protection of Human Subjects at the University of California at Berkeley, by committees on the use of human subjects at each of the participating hospitals, and by an external Safety and Data Monitoring Committee appointed by the National Heart, Lung, and Blood Institute.
The RR program combined lipid-altering therapies with counseling and training in the modification of diet, exercise, and other lifestyle factors. Major lipid-altering goals of RR were decrease in LDL-C levels to <2.84 mmol/L (110 mg/dL), decrease in TG to <1.13 mmol/L (100 mg/dL), and increase in HDL-C to >1.42 mmol/L (55 mg/dL); for RR subjects considered unlikely to reach the LDL-C goal within the first year of the study, a cholesterol-lowering drug was added to the intervention on lifestyle. Drug therapies included bile acid–binding resins, niacin, HMG-CoA reductase inhibitors (lovastatin, available only during the last 2 years of the study), and fibrates.1
Computer-assisted coronary quantification was performed at baseline and after 4 years. For each segment, changes in minimum diameter, mean diameter, and percent stenosis were calculated. The annualized rate of change in minimum diameter of coronary artery segments with angiographically visible atherosclerotic lesions at baseline, averaged over all such segments in each subject, is denoted by ΔMin (mm/y). ΔMin was the primary measure of CAD progression. For the 246 subjects who completed SCRIP, ΔMin was −0.045±0.006 mm/y in UC (n=127) and −0.024±0.006 mm/y in RR (n=119, P=.02)1 ; negative signs indicate arterial narrowing. For the 213 men studied in this report, most of the segments quantified would be considered moderately rather than severely stenosed: at baseline, median stenosis for these 666 segments was 30%, and >95% of the segments were <50% stenosed.
Limitations imposed by the SCRIP design on the assessment of the impact of any one therapy on the outcome have been discussed elsewhere2 ; they include the use of multiple drugs (often concurrent) for many subjects, variations in dose and duration of use for each drug, and variations in lifestyle interventions. Taken together, these and other aspects of the multifactorial, individualized treatment for both RR and UC subjects imply low statistical power to isolate the contribution of the effect of any specific therapy on the SCRIP outcome.
All subjects had their clinical status and cardiac risk factors evaluated at baseline before randomization and annually for 4 years. Baseline clinic visits occurred at least 3 weeks after hospital discharge, after entry coronary angiography. Staff collecting data in the clinic were not blinded to group assignment of subjects, but staff performing laboratory measurements were blinded. At baseline and annually, fasting plasma lipids were measured at two clinic visits, usually within 2 weeks of each other; averages of the two values were used to represent subjects. Total cholesterol, LDL-C, HDL-C, and TG were measured in the Stanford Lipoprotein Research Laboratory as described previously.1 Blood samples were collected in Na2EDTA (1.5 mg/mL), and plasma was transported on wet ice to Donner Laboratory, for analyses by ANUC and GGE and for measurement in a subset of subjects of apoB by standardized ELISA1 21 using two monoclonal capture antibodies (Medix Biotech).
ANUC and Determination of Density Mode
ANUC18 measures of LDL mass concentrations at Donner Laboratory were made in 11 intervals of Sof. These intervals have been grouped into four subclasses13 : Sof 0-3 (LDL IV) and Sof 3-5 (LDL III), whose sum is considered to estimate the concentration of small dense LDL (Sof 0-5); and Sof 5-7 (LDL II) and Sof 7-12 (LDL I), whose sum is considered to estimate the concentration of large buoyant LDL (Sof 5-12). IDL are estimated by Sof 12-20, and VLDL are estimated by Sof 20-400. Similarly, ANUC was used to measure HDL mass concentrations in two major subgroups with flotation rates of 0 to 3.5 (HDL3) and 3.5 to 9 (HDL2).
From values of peak flotation rates, ANUC also calculates a buoyant density, σ, that characterizes flotation peaks of LDL profiles.18 Accuracy of peak σ is estimated to be 0.001 g/mL. A strong correspondence between σ (also called hydrated density) and LDL density determined by standard preparative ultracentrifugation has been established for specific LDL density subfractions,22 and this correspondence is used to estimate a preparative density corresponding to a particular value of peak σ.18
Fig 1⇓ shows the baseline frequency distribution of σ, and corresponding preparative density values, for the 213 SCRIP subjects discussed in this report. There are two large modes in this distribution; we refer to them as density modes. We have found a very similar bimodality in CAD subjects in MARS (B.D. Miller, R.M. Krauss, H.H. Hodis, unpublished data, 1995) and in 177 healthy subjects on low-fat diets (B.D. Miller, R.M. Krauss, D.M. Dreon, unpublished data, 1995). Fig 1⇓ also shows the distribution of σ for the combined population consisting of the 177 healthy subjects together with all 291 SCRIP subjects and all 263 MARS subjects for whom σ was measured at baseline. On the basis of data from all of these studies, we chose the cut point σ=1.0325 g/mL to define the two modes in the distribution. LDL particle profiles with σ ≤1.0325 g/mL and σ >1.0325 g/mL are called buoyant-mode and dense-mode profiles, respectively. Preparative density corresponding to σ=1.0325 g/mL is ≈1.0378 g/mL.22
GGE and Determination of LDL Subclass Pattern
GGE was performed using Pharmacia PPA 2/16% gradient gels, as described previously.13 15 Stained gels were scanned with a Transidyne RFT Scanning Densitometer, and peak particle sizes were calculated from calibration curves using standards of known size. Accuracy of peak particle size is estimated to be 3 Å. Resulting scans were assigned an LDL subclass pattern, as described previously.13 15 Pattern A is characterized by a major peak size of ≥264 Å, often with skewing toward smaller particles. Pattern B is characterized by a major peak size of ≤255 Å, often with skewing toward larger particles. Intermediate profiles, designated pattern I, typically have a single or double peak in the range of 256 to 263 Å.
To determine whether two or more distinct groups of subjects differed in their response to intervention, we compared results in RR with results in UC separately for each of the groups. Comparisons of RR with UC were made with Wilcoxon tests for TG and for all angiographic and ANUC measurements because distributions for many of those measurements (including ΔMin) are nonnormal. For all other variables, comparisons of RR to UC were made with t test. Correlations reported are Spearman correlations. Calculations were performed using SAS software. Mean values are expressed as mean±SEM for all available data. (For some variables in some subjects, values were missing.) We use the term “significant” to refer to results with a value of P<.01.
In performing multiple comparisons between two groups, we did not apply a Bonferroni correction to the P values that were generated. This is primarily because in attempting to identify variables that may provide insight into our primary result (a result not subject to Bonferroni concerns because it involves the primary SCRIP outcome variable), we considered it preferable to identify such variables by a broad rather than a narrow criterion.
Analyses by Density Mode
Table 1⇓ shows a comparison of clinical, laboratory, and angiographic data at baseline between the 121 buoyant-mode and the 92 dense-mode subjects in SCRIP. As expected on the basis of definitions of the modes, buoyant-mode subjects had much higher levels of large, buoyant LDL than did dense-mode subjects; lower levels of small, dense LDL; lower peak density; and higher peak flotation rate. Further significant differences between the modes were those in levels of TG, IDL, and VLDL, which were higher in dense-mode subjects, and in HDL-C and HDL2, which were higher in buoyant-mode subjects. Although small, dense LDL has been reported to be a risk factor for CAD, severity of disease in arteries was not significantly worse in dense-mode subjects. However, because subject entry criteria included symptoms of CAD, SCRIP does not allow assessment of the relation of small, dense LDL to the risk of CAD.
Table 2⇓ shows the annualized rates of change of angiographic variables. For buoyant-mode subjects, intervention had no effect: ΔMin was −0.038±0.007 mm/y in UC and −0.039±0.010 mm/y in RR. In contrast, dense-mode subjects showed a significant benefit of intervention: ΔMin was −0.054±0.012 mm/y in UC and −0.008±0.009 mm/y in RR (P=.007).
Table 3⇓ shows 4-year changes for risk factors. Differences in LDL-C changes between UC and RR were similar and comparably significant in the two modes. Differences between UC and RR in TG changes were substantial for subjects in both modes, although notably larger in dense mode; these larger differences are not surprising, in view of the higher baseline TG levels of dense-mode subjects together with the TG-lowering component of the RR program. Subjects in both modes showed large differences between UC and RR in LDL subclass changes. Also, for both modes, RR compared with UC produced increases in levels of HDL-C and HDL2; decreases in levels of apoB, VLDL, and IDL; and decreases in body mass index.
Changes in density mode between baseline and the end of the study occurred for 19% of buoyant-mode subjects (24% in UC and 14% in RR) and 29% of dense-mode subjects (31% in UC and 28% in RR). Angiographic response to intervention was similar for those who did and those who did not change modes (data not shown).
We looked for relationships between density mode, angiographic changes, and changes in risk factors that would suggest explanations for the difference between buoyant-mode and dense-mode subjects in angiographic response to RR. In particular, we considered the possibility that there were changes in specific risk factors that correlated with angiographic changes and that were more strongly affected by RR for dense-mode than for buoyant-mode subjects. However, our exploratory analyses did not find any risk factors meeting these criteria.
We looked for baseline differences between UC and RR that were large enough (P<.15) to warrant the possibility of adjusting ΔMin for those differences. For buoyant-mode subjects, there were differences in HDL2 and VLDL; however, the differences would be considered advantageous for RR subjects, who had higher HDL2 and lower VLDL levels than UC subjects. Thus, we would not expect adjustment to alter the finding that buoyant-mode subjects did not benefit from RR. For dense-mode subjects, although UC and RR differed in percent stenosis (P=.09) and minimum diameter (P=.11), ΔMin did not correlate with either of these variables, so we did not adjust ΔMin.
We considered the possibility that the favorable response to RR by dense-mode subjects but not buoyant-mode subjects was due to differences in the use of particular lipid-altering drugs. Although a definitive analysis of this possibility is precluded by the SCRIP design, we did attempt to evaluate the possibility that disproportionate use of a particular drug by buoyant-mode and dense-mode subjects accounted for the different response to RR by these two groups. We followed the approach previously used to estimate the effect of a single drug on the outcome for all subjects.2 In this approach, a subject is categorized as a user or nonuser of a specific drug if the subject did or did not, respectively, use that drug consistently during the last 2 years of the study. By this criterion of drug use, resin use among the 213 men studied here was 69% in RR (68% in buoyant-mode and 71% in dense-mode) and 8% in UC (both modes). Corresponding values for use of niacin were 28% in RR (20% in buoyant-mode and 39% in dense-mode) and 8% in UC (6% and 10%); for use of lovastatin, values were 21% in RR (14% in buoyant-mode and 29% in dense-mode) and 6% in UC (both modes); and for use of fibrates, values were 18% in RR (5% in buoyant-mode and 34% in dense-mode) and 3% in UC (3% and 4%).
Niacin, lovastatin, and fibrates were used in RR disproportionately by dense-mode subjects. However, the small number of subjects in either mode who used each of these drugs suggests it is unlikely that any one of these drugs was a large factor in our results. To verify this suggestion, for each drug separately, we repeated the separate analyses of ΔMin for each mode, after first excluding from the 213 subjects all who were categorized as users of that drug. Angiographic results in all three cases were very similar to those for the larger groups of subjects in both modes (data not shown), so we concluded that the preferential benefit of RR for dense-mode subjects was not the result of disproportionate use of niacin, lovastatin, or fibrates.
Resins were used by very similar proportions of buoyant-mode and dense-mode subjects, so it does not appear that the preferential benefit of intervention for dense-mode subjects was due to disproportionate use of resins. However, another possible role of resins in our results emerges from a comparison of ΔMin for resin users in RR with ΔMin for nonusers in UC, separately for each mode (ie, from analyses that exclude UC subjects who used resins and RR subjects who did not). For buoyant-mode subjects, ΔMin was −0.040±0.008 mm/y in the UC subjects who did not use resins (n=60) and −0.038±0.014 mm/y in the RR subjects who did (n=38, P=.45). For dense-mode subjects, ΔMin was −0.056±0.013 mm/y in the UC subjects who did not use resins (n=47) and −0.007±0.007 mm/y in the RR subjects who did (n=29, P=.006). These comparisons, similar to those for the larger groups of buoyant-mode and dense-mode subjects, suggest a possible contributing factor for our overall results: resins have a relatively strong effect on atherogenic processes in dense-mode subjects compared with those in buoyant-mode subjects.
This suggestion regarding resins follows from one particular definition of resin users and nonusers; it does not account for variations in dose or duration of use, concurrent use of other drugs, and other factors that would be expected to affect outcome. We are compiling a detailed database on drug use that will provide the basis for more rigorous analyses than are presently possible of the effect of lipid-altering therapies on angiographic outcome in SCRIP.
We considered the possibility that σ considered as a continuous rather than a bimodal variable was a better predictor of angiographic benefit of RR. Our exploratory analyses included separate comparisons of UC and RR for each tertile of σ (Table 4⇓). These results suggest that RR was not beneficial to subjects in the lowest tertile but that it was beneficial to subjects in both the middle and highest tertiles. However, subdivision of the middle tertile into its constituent 50 buoyant-mode and 21 dense-mode subjects leads to a different view. For these buoyant-mode subjects, ΔMin was −0.031±0.009 mm/y in UC (n=31) and −0.020±0.013 mm/y in RR (n=19, P=.39), which suggests a better response to RR for these subjects than for subjects in the lowest density tertile. However, these 50 buoyant-mode subjects appeared to be less responsive to RR than dense-mode subjects in the middle tertile, for whom ΔMin was −0.092±0.026 mm/y in UC (n=13) and −0.017±0.016 mm/y in RR (n=8, P=.06). The small sample sizes used do not permit discrimination between the possibility that the different response by buoyant-mode and dense-mode subjects occurred because of reasons related to mode as such and the possibility that the difference reflects a difference in σ viewed as a continuous parameter.
Subjects in the two density modes showed large, significant differences in levels of several lipid and lipoprotein variables (Table 1⇑). These variables all correlate very significantly (P<.0001) with σ: for TG; HDL-C; HDL2; IDL; VLDL; small, dense LDL; and large, buoyant LDL, the correlations with σ are 0.71, −0.51, −0.54, 0.37, 0.75, 0.83, and −0.50, respectively. For some of these variables—notably the levels of small, dense LDL and of large, buoyant LDL—large differences between the modes are expected from the definition of the modes. For the other variables (TG, HDL-C, HDL2, IDL, VLDL), such differences are not directly implied by the definition of the modes, and reasons for the strong statistical relationships of these variables to LDL particle profiles are not established.17
The possible use of TG as an indicator of density mode is illustrated in Fig 2⇓, which shows TG distributions separately for buoyant-mode and dense-mode subjects. Fig 2⇓ shows that for either relatively low or relatively high levels, TG is a strong indicator of density mode. For example, >90% of the subjects with TG <1.13 mmol/L (100 mg/dL) were buoyant-mode subjects and >90% of those with TG >1.81 mmol/L (160 mg/dL) were dense-mode subjects. In the intermediate range, 1.13 to 1.81 mmol/L (100 to 160 mg/dL), although the majority of subjects were in the buoyant mode, TG was not a strong indicator of mode.
For TG and all other variables that differed significantly between modes, we explored the ability of those variables to predict angiographic benefit of RR. To do this, we dichotomized subjects into those who had relatively high and relatively low levels, as defined by medians of these variables, and then compared UC with RR separately for each group of subjects (Table 5⇓). Although medians need not be optimal cut points for defining “relatively high” or “relatively low,” the distributions of these variables—unlike that of σ—did not strongly suggest other cut points. Table 5⇓ shows that several variables that differ strongly between density modes predicted, comparably well to density mode, the angiographic benefit of RR.
Analyses by LDL Subclass Pattern
Table 6⇓ shows the correspondence between LDL subclass pattern as assessed by GGE and the classification by density mode for the 213 subjects that we discuss. As is to be expected from the high correlation between size and density of LDL particles (r=−.8 in SCRIP), most pattern A subjects were buoyant-mode subjects, and there was a large overlap between pattern B subjects and dense-mode subjects. Pattern I subjects, who have no analogue in the density-based classification, were predominantly buoyant-mode subjects.
Table 7⇓ compares ΔMin in UC and RR for the three types of LDL subclass pattern. Although the difference between UC and RR was not significant for any of the patterns, results for pattern A and pattern B subjects were, as would be expected from Table 6⇑, similar to results for buoyant-mode and dense-mode subjects, respectively.
Although mean results for pattern I and pattern B subjects appear similar, the benefit of RR seen for pattern I subjects is not indicative of pattern I subjects in general but rather was due primarily to the 10 dense-mode subjects among the pattern I subjects (data not shown).
We report that the goal of RR in SCRIP, reduction of CAD progression as determined by rates of change of minimum artery diameter, was achieved in male subjects whose baseline LDL profile, as determined by ANUC, peaked in small, dense LDL. Reduction in CAD progression was not seen in subjects whose profile peaked in larger, more buoyant LDL. The two types of profile were defined by a bimodality in peak LDL density.
All of the subjects who we analyzed were male patients with CAD; our results should not be assumed to extend to women or to subjects without symptoms of CAD. Furthermore, because most of the arteries analyzed were those that would be considered moderately rather than severely diseased, our results should not be assumed to extend to patients with more severely diseased arteries. However, in view of reports that lesions resulting in cardiac events are typically only moderately diseased,23 24 25 26 a benefit of therapy for moderately stenosed lesions, such as that described here, may be of greater clinical import than a benefit for severely stenosed lesions.
Although our primary categorization of subjects was defined by a bimodality in peak LDL density (Fig 1⇑), dichotomies defined by relatively high or relatively low levels of lipid and lipoprotein measurements that strongly correlated with σ (notably TG and HDL-C) predicted benefit of RR comparably well to the bimodality. The strong correlations and the study design preclude determination of whether any one of these variables was substantially better than the others as a predictor of angiographic benefit of RR. Because we consider it likely that the two density modes result from distinct sets of processes regulating lipoprotein metabolism,17 we think it plausible that those two modes may be the best markers of the metabolic differences between subjects who responded favorably to RR and those who did not.
Although we focused on the bimodality in peak LDL density, the multiplicity of distinct LDL components seen with the use of techniques other than ANUC (notably, GGE and density gradient ultracentrifugation)13 suggests that each of the two large density modes in Fig 1⇑ represents a superposition of LDL subspecies that are not distinguished on the basis of σ alone. Such subspecies may have distinct metabolic origins, and for that reason, among others, one might expect the relationship between density and response to intervention in SCRIP to be more complex than can be described simply by the bimodality. That expectation finds some support in the fact that buoyant-mode subjects in the middle tertile of σ (ie, subjects with “moderately” buoyant LDL) showed somewhat better response to RR than subjects in the lowest tertile, who did not show even a tendency toward benefit of RR. Evidently, effective treatment of CAD for men with “very” buoyant LDL may require therapies other than those used extensively in RR. HMG-CoA reductase inhibitors, which were not widely used in SCRIP, have been reported to be beneficial in treating CAD7 8 9 and in reducing the incidence of cardiac events.27 28 However, preliminary analyses of MARS on the basis of tertiles of σ found no angiographic benefit of lovastatin for men in either the lowest or the highest tertile of σ; in contrast, a strong benefit was seen for men whose LDL profiles peaked in the middle density tertile.29
Our present report also includes angiographic results for SCRIP subjects categorized by LDL subclass pattern, as determined by GGE. This size-based classification of LDL profiles did not predict angiographic benefit of RR as strongly as the density-based classification. This may be because particle density discriminates small, dense LDL from other LDL subspecies more precisely than particle size. It is also possible that LDL density is more representative than LDL size of processes that govern metabolic responsiveness to particular therapies.
Bile acid–binding resins were used much more extensively than any other drug in SCRIP (in both modes). Thus, one possible explanation for our findings is that resins have substantially more effect on atherogenic processes in dense-mode subjects than in buoyant-mode subjects. Our preliminary assessments of the role of resins in our results suggest that this is so.
This suggestion on the basis of SCRIP data is confounded by factors related to study design. However, it is consistent with results of a recent analysis30 of CLAS,10 11 an angiographic trial of the effects of resins plus niacin on CAD progression. That analysis found that subjects above the TG median showed very significant angiographic benefit of therapy, whereas those below showed no effect. Although the differing use of niacin in SCRIP and CLAS confounds a comparison of the two studies, the similarity of their results on the basis of TG suggests that the same metabolic factors may be underlying both CLAS and SCRIP results. In particular, our conjecture that metabolic processes underlying density mode played a critical role in SCRIP results also applies to CLAS.
SCRIP results may also be related to those in FATS,4 in which a subgroup of subjects with relatively low HDL-C and relatively high TG, levels that suggest that these subjects would be classified as dense-mode subjects, was found to be more responsive to intensive lipid-lowering therapy than subjects with higher levels of HDL-C and lower levels of TG but higher LDL-C. In FATS, however, as in SCRIP, the use of multiple therapies confounds the analysis of questions as to whether specific therapies were preferentially beneficial to subjects with particular lipid or lipoprotein profiles.
Metabolic studies may contribute to an explanation of the apparent capability of lipid and lipoprotein profiles to predict response to specific lipid-altering therapies. For example, the combination of colestipol and simvastatin has been reported to have different effects on apoB metabolism than colestipol alone.31 Such differences might be expected on the basis of the different cholesterol-lowering mechanisms of HMG-CoA reductase inhibitors and bile acid–binding resins.
At present, we have no basis for a judgment as to whether peak LDL density or one of its correlates is the most informative regarding physiological processes that determined benefit of RR in SCRIP. Even so, the fact that there were baseline variables that predicted angiographic benefit is in itself important; more generally, it is important that there is now evidence from several studies that the efficacy of CAD therapies is not uniform across the population of CAD patients. Although specific therapies differed in these studies, the cumulative results suggest, either directly (in cases where LDL density was measured) or indirectly (via the correlations of density to TG and HDL-C), that subjects with different types of LDL particle profile also differ in their response to CAD therapies. We find it plausible that a common basis for different types of LDL profile and for different responses to therapies is to be found at the level of underlying metabolic processes. That is, we expect that distinct sets of metabolic processes may give rise both to different types of LDL profile and to different responses to a specific lipid-altering therapy. If so, then LDL profiles may be useful indicators of optimal therapy for individual patients.
Selected Abbreviations and Acronyms
|CAD||=||coronary artery disease|
|CLAS||=||Cholesterol-Lowering Atherosclerosis Study|
|FATS||=||Familial Atherosclerosis Treatment Study|
|GGE||=||gradient gel electrophoresis|
|MARS||=||Monitored Atherosclerosis Regression Study|
|Sof||=||Svedberg flotation rate|
|SCRIP||=||Stanford Coronary Risk Intervention Project|
|UC||=||usual care of physicians|
This work was supported by grants RO1-HL-33577 and HL-18574 from the National Heart, Lung, and Blood Institute and was conducted at the E.O. Lawrence Berkeley National Laboratory through the US Department of Energy under contract DE-AC03-76SF00098. We thank Joseph Orr for carrying out the ANUC measurements and Dennis Duncan and Laura Holl for performing GGE. We also thank Howard Hodis for permission to use data obtained from the Monitored Atherosclerosis Regression Study.
Reprint requests to Ronald M. Krauss, MD, E.O. Lawrence Berkeley National Laboratory, University of California, Donner Laboratory, Room 465, One Cyclotron Rd, Berkeley, CA 94720. E-mail firstname.lastname@example.org.
- Received January 9, 1996.
- Revision received May 20, 1996.
- Accepted May 27, 1996.
- Copyright © 1996 by American Heart Association
Haskell WL, Alderman EL, Fair JM, Maron DJ, Mackey SF, Superko HR, Williams PT, Johnstone IM, Champagne MA, Krauss RM, Farquhar JW. Effects of intensive multiple risk factor reduction on coronary atherosclerosis and clinical cardiac events in men and women with coronary artery disease: the Stanford Coronary Risk Intervention Project (SCRIP). Circulation. 1994;89:975-990.
Alderman EL, McMillan A, Haskell W. Effects of the lipid intervention in the Stanford Coronary Risk Intervention Project. In: Bruschke AVG, Reiber JHC, Lie KI, Wellens HJ, eds. Lipid-Lowering Therapy and Progression of Atherosclerosis. Dordrecht, Netherlands: Kluwer Academic Publishers; 1996.
Watts GF, Lewis B, Brunt JNH, Lewis ES, Coltart DJ, Smith LDR, Mann JI, Swan AV. Effects on coronary artery disease of lipid-lowering diet, or diet plus cholestyramine, in the St Thomas' Atherosclerosis Regression Study (STARS). Lancet. 1992;1:563-569.
Blankenhorn DH, Azen SP, Kramsch DM, Mack WJ, Cashin-Hemphill L, Hodis HN, DeBoer LWV, Mahrer PR, Masteller MJ, Vailas LI, Alaupovic P, Hirsch LJ, MARS Research Group. Coronary angiographic changes with lovastatin therapy: the Monitored Atherosclerosis Regression Study (MARS). Ann Intern Med. 1993;119:969-976.
Hodis HN, Mack WJ, Azen SP, Alaupovic P, Pogoda JM, LaBree L, Hemphill LC, Kramsch DM, Blankenhorn DH. Triglyceride- and cholesterol-rich lipoproteins have a differential effect on mild and moderate/severe lesion progression as assessed by quantitative coronary angiography in a controlled trial of lovastatin. Circulation. 1994;90:42-49.
Waters D, Higginson L, Gladstone P, Kimball B, Le May M, Boccuzzi SJ, Lespe´rance J, CCAIT Study Group. Effects of monotherapy with an HMG-CoA reductase inhibitor on the progression of coronary atherosclerosis as assessed by serial quantitative arteriography: the Canadian Coronary Atherosclerosis Intervention Trial. Circulation. 1994;89:959-968.
Blankenhorn DH, Alaupovic P, Wickham E, Chin HP, Azen SP. Prediction of angiographic change in native human coronary arteries and aortocoronary bypass grafts: lipid and nonlipid factors. Circulation. 1990;81:470-476.
Ornish D, Brown SE, Scherwitz LW, Billings JH, Armstrong WT, Ports TA, McLanahan SM, Kirkeeide RL, Brand RJ, Gould KL. Can lifestyle changes reverse coronary heart disease? Lancet. 1990;1:129-133.
Griffin BA, Freeman DJ, Tait GW, Thomson J, Caslake MJ, Packard CJ, Shepherd J. Role of plasma triglyceride in the regulation of plasma low density lipoprotein (LDL) subfractions: relative contribution of small, dense LDL to coronary heart disease risk. Atherosclerosis. 1994;106:241-253.
Lindgren FT, Jensen LC, Hatch FT. The isolation and quantitative analysis of serum lipoproteins. In: Nelson GJ, ed. Blood Lipids and Lipoproteins: Quantitation, Composition, and Metabolism. New York, NY: John Wiley and Sons; 1972:181-274.
McNamara JR, Campos H, Ordovas JM, Peterson J, Wilson PWF, Schaefer EJ. Effect of gender, age, and lipid status on low density lipoprotein subfraction distribution: results from the Framingham Offspring Study. Arteriosclerosis. 1987;7:483-490.
Williams PT, Vranizan KM, Krauss RM. Correlations of plasma lipoproteins with LDL subfractions by particle size in men and women. J Lipid Res. 1992;33:765-774.
Burdon RH, Van Knippenberg PH, eds. Practice and Theory of Enzyme Immunoassays. Amsterdam, Holland: Elsevier Science Publishers; 1985:345-347; series Laboratory Techniques in Biochemistry and Molecular Biology.
Little WC, Constantinescu M, Applegate RJ, Kutcher MA, Burrows MT, Kahl FR, Santamore WP. Can coronary angiography predict the site of a subsequent myocardial infarction in patients with mild-to-moderate coronary artery disease? Circulation. 1988;78:1157-1166.
Brown BG, Gallery CA, Badger RS, Kennedy JW, Mathey D, Bolson EL, Dodge HT. Incomplete lysis of thrombus in the moderate underlying atherosclerotic lesion during intracoronary infusion of streptokinase for acute myocardial infarction: quantitative angiographic observations. Circulation. 1986;73:653-661.
Scandinavian Simvastatin Survival Study Group. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet. 1994;1:563-589.
Miller BD, Cashin-Hemphill L, Mack WJ, Hodis HN, Krauss RM. Predominance of mid-density low density lipoproteins predicts angiographic benefit of lovastatin in the Monitored Atherosclerosis Regression Study. Circulation. 1994;90(suppl I):I-460. Abstract.
Miller BD, Krauss RM, Cashin-Hemphill L, Blankenhorn DH. Baseline triglyceride levels predict angiographic benefit of colestipol plus niacin therapy in the Cholesterol-Lowering Atherosclerosis Study (CLAS). Circulation. 1993;4(suppl I):I-363. Abstract.
Gaw A, Packard CJ, Lindsay GM, Murray EF, Griffin B, Caslake MJ, Colquhoun I, Wheatley DJ, Lorimer AR, Shepherd J. Effects of colestipol alone and in combination with simvastatin on apolipoprotein B metabolism. Arterioscler Thromb Vasc Biol. 1996;16:236-249.