HDL Cholesterol Level Predicts Survival in Men After Coronary Artery Bypass Graft Surgery
20-Year Experience From The Cleveland Clinic Foundation
Background—HDL cholesterol (HDL-C) is an important independent predictor of atherosclerosis, yet the role that HDL-C may play in the prediction of long-term survival after CABG remains unclear. The risk associated with a low HDL-C level in post-CABG men has not been delineated in relation to traditional surgical variables such as the use of arterial conduits, left ventricular function, and extent of disease.
Methods and Results—We performed a prospective, observational study of 432 men who underwent CABG between 1978 and 1979 in whom preoperative HDL-C values were available. Baseline lipid and lipoprotein values, history of diabetes mellitus and hypertension, left ventricular ejection fraction, extent of disease, and use of internal thoracic arteries were recorded. Hazard ratios (HRs) were determined in the patients with and without a low HDL-C level, which was defined as the lowest HDL-C quartile (HDL-C ≤35 mg/dL). After adjustment for age, as well as for baseline metabolic parameters and surgical variables just noted, HDL-C corresponded to both overall (HR 0.40, CI 0.20 to 0.83, P=0.01) and event-free (HR 0.41, CI 0.24 to 0.70, P=0.001) survival. Patients with a high HDL-C level (>35 mg/dL) were 50% more likely to survive at 15 years than were patients with low HDL-C level (≤35 mg/dL) (74% versus 57% adjusted survival, respectively; HR 1.72, P=0.005). In addition, HDL-C showed a strong effect on time-to-event survival such that patients with an HDL-C level of >35 mg/dL were 50% more likely to survive without a subsequent myocardial infarction or revascularization (HR 1.42, P=0.02).
Conclusions—HDL-C is an important predictor of survival in post-CABG patients. In this study of >8500 patient-years of follow-up, HDL-C was the most important metabolic predictor of post-CABG survival. One third fewer patients survive at 15 years if their HDL-C levels are ≤35 mg/dL at the time of CABG. The measurement of HDL-C provides a compelling strategy for the identification of high-risk subsets of patients who undergo CABG.
The inverse relation between HDL cholesterol (HDL-C) and coronary artery disease (CAD) events has been established in many epidemiological studies.1 2 3 4 Although the importance of HDL as an independent predictor of atherosclerosis has been reported, its role in the prediction of survival after CABG is less certain. The mortality risk associated with a low HDL-C level in post-CABG men has not been delineated in relation to traditional surgical variables such as use of arterial conduits, left ventricular (LV) function (LV ejection fraction [LVEF]), and extent of disease.
To address these concerns, we chose to analyze the effects of a low HDL-C level on the mortality risk associated with CABG in a population-based cohort. We identified a cohort of men enrolled in the Cardiovascular Information Registry at the Cleveland Clinic Foundation who underwent CABG between 1978 and 1979 and had a full lipid assessment. This is a well-characterized population of patients who underwent primary elective myocardial revascularization and were followed for >15 years.
The Cardiovascular Information Registry (CVIR)5 was initiated at the Cleveland Clinic in the early 1970s, with follow-up of patient survival and events at an interval basis. All descriptive terms have previously been described. Follow-up was conducted for all 432 patients. Routine follow-up was scheduled at 5-year intervals for these patients. In addition, chart review was conducted to determine whether further events had been recorded in the Cleveland Clinic medical record. Finally, a search of the Social Security Administration database was conducted to verify vital status.
For this study, 2244 patients underwent cardiovascular revascularization between 1978 and 1979. Of these, subjects were excluded if they were undergoing a repeat CABG or combined valvular procedure (n=313) or died before hospital discharge (n=11). Women were excluded due to the small number (n=239). Of the remainder, patients without a documented total cholesterol (TC; n=5), triglycerides (TG; n=390), or HDL-C (n=834) level were excluded. Complete lipid data were available for 432 men. Major exclusions are given in Table 1⇓.
Baseline lipid and lipoprotein values, history of diabetes mellitus (DM) and hypertension (HTN), extent of disease, and use of internal thoracic arteries (ITAs) were recorded. Hazard ratios (HRs) were determined in the patients with and without a low HDL-C level, which was defined as the lowest HDL-C quartile (HDL-C ≤35 mg/dL).
Fasting TC, HDL-C, and TG concentrations were considered to be baseline values if blood was drawn within 30 days before CABG. Lipoproteins were measured in serum after an overnight fast in our hospital-based Centers for Disease Control and Prevention (CDC)-standardized laboratory according to Lipid Research Clinics methodology.6
Time-to-event data were analyzed according to the primary end point of all-cause mortality (overall survival). Cause-specific mortality was unavailable through this registry. However, all-cause mortality is thought to be an objective and unbiased end point.7 8 Mortality was quantified beyond 30 days after surgery (“late mortality”) to avoid the influence of immediate postoperative complications. The secondary end point of overall survival free from myocardial infarction (MI) or repeat revascularization (event-free survival) was analyzed. Event-free survival models were developed based on time to the earliest event. Survival estimates were generated with the Kaplan-Meier method.9 Cox proportional hazards modeling (PROC PHREG; SAS Institute) was used to assess the relative importance of baseline risk factors to the end points.10 Tied event data were handled with Efron’s method of approximation.11 Overall model significance was assessed with likelihood-ratio tests, and the significance of each variable in the model was assessed with the Wald test. HRs are presented (with 95% CIs) to show the risk of an event when the factor is present.
The primary analyses involved 2 steps. Overall and event-free survival rates were first evaluated with continuous measures of lipid parameters (with natural log transformations in consideration of normality assumptions) and then evaluated with categorical variables based on quartile distributions (high-risk quartile versus other quartiles). To eliminate the multiple collinearity of HDL-C and TG, we developed a model with TG used as a stratification variable (stratified by quartiles). All analyses were performed with Version 6.12 of the SAS statistical package (SAS Institute).
Baseline characteristics for the total cohort on the basis of HDL-C availability are included in Table 2⇓. Surviving subjects were followed for a median of 16.3 years. Marked consistency of results was noted between groups. The 15-year survival rates for the 2 groups were similar (67±2.3% with measured HDL-C, 67±1.7% without measured HDL-C; mean±SEM). These presentation characteristics are notably different from those of other reported populations. Our cohort, although in general younger, less obese, less often hypertensive, and less often diabetic than the current characterization of our surgical population, is representative of patients who underwent CABG in the late 1970s. Of significance, our cohort had a high percentage of ITA use (40%). Baseline characteristics of patients with low (≤35 mg/dL) or high (>35 mg/dL) measured HDL-C levels are included in Table 3⇓.
The relationships between TC, TG, and HDL-C and all-cause mortality were first examined. There were 157 total deaths in the population. HDL-C was the sole lipid parameter that showed a significant relationship to survival (Table 4⇓). The HR of 0.40 (CI 0.20 to 0.83, P=0.01) suggests that the mortality rate is decreased 3% for each 1-mg/dL increase in HDL-C level. Neither TG (P=0.48) nor TC (P=0.12) displayed significant relationships to overall survival. Age (HR 1.03, CI 1.01 to 1.06, P=0.007) and ITA use (HR 0.34, CI 0.22 to 0.51, P=0.0001) were the only other covariates significantly related to overall survival. An interaction between HDL-C and ITA use was not significant when entered into the model (P=0.31), suggesting that the influence of HDL-C on survival was not dependent on the presence or absence of an ITA.
Using cut points based on our quartile analysis, we note that patients with a high HDL-C level (>35 mg/dL) were 50% more likely to survive at 15 years than were patients with a low HDL-C level (<35 mg/dL) (74% versus 57% adjusted survival, respectively; HR 1.72, P=0.005). The Kaplan-Meier analyses for overall survival (Figure 1⇓) show that the low and high HDL-C curves begin to diverge soon after surgery. However, in this cohort, the difference does not reach statistical significance until the sixth or seventh year after surgery.
Consistent with overall survival, HDL-C predicted survival free from MI or repeat revascularization with an HR of 0.41 (CI 0.24 to 0.70, P=0.001) (Table 5⇓). ITA use (HR 0.60, CI 0.46 to 0.78, P=0.0001) was the only other significant predictor of event-free survival time, although the effect of HTN was borderline (HR 1.50, CI 0.97 to 2.34, P=0.07). Again, there was no evidence of a significant interaction among ITA, HDL-C, and event-free survival (P=0.28).
Patients with an HDL-C level of >35 mg/dL were, again, 50% more likely to survive without a subsequent MI or revascularization than were those with an HDL-C level of ≤35 mg/dL (45% versus 30% adjusted event-free survival, respectively; HR 1.42, P=0.02). TC and TG remained nonsignificant (P=0.45 and P=0.57. respectively). Again, the Kaplan-Meier analyses for event-free survival (Figure 2⇓) show early divergence of these curves between the low and high HDL-C groups that reaches statistical significance at the sixth or seventh year after surgery.
HDL-C and TG are known to be strongly associated. To control for TG as a confounder, we developed a model with TG stratified by quartiles. HDL-C continues to show a significant effect with regard to both overall and event-free survival (HR 0.37, 95% CI 0.18 to 0.77, P=0.008; HR 0.41, 95% CI 0.24 to 0.69, P=0.001, respectively).
We demonstrate that baseline HDL-C level dramatically predicts overall and event-free survival in men who underwent CABG and were followed for up to 20 years after surgery. This remained true even after adjustment for DM, body mass index, and TG, as well as surgical variables, including graft type, ventricular function, and extent of disease. In our population, patients with a high HDL-C level (>35 mg/dL) have a 50% greater overall survival than do patients with a low HDL-C level (≤35 mg/dL).
Epidemiological studies have demonstrated the significant inverse relationship between HDL-C and coronary atherosclerosis. The Framingham Heart Study revealed HDL-C to have the strongest relation (inverse) to CAD of all standard lipoproteins measured.12 13 14 15 16 In general, a 1-mg/dL increase in HDL-C level is associated with a significant reduction in CAD risk of 2% in men and of 3% in women.13 The estimate in our current analysis is 3% in men.
Several studies also support the role of HDL-C in the prediction of death and cardiovascular events in patients with CAD. In the Lipid Research Clinics Coronary Primary Prevention Trial (CPPT),17 the TC/HDL-C ratio added risk-discriminating ability to TC and LDL-cholesterol (LDL-C) measures alone. In the Tromso Heart Study,18 >6000 young men (20 to 49 years old) were followed for 2 years. In this study, HDL-C provided a 3-fold greater contribution to the prediction of CHD events than did LDL-C in this population (RR 1.063). The mean HDL-C level was 25.6 mg/dL in the MI patients compared with 39.4 mg/dL in control subjects (P<0.005). Recent data from the VA HDL Intervention Trial (VA-HIT) study19 suggest that for a 1-mg/dL increase in HDL-C level, there was a ≈3% decrease in the primary end point of CHD death or MI, a 3% reduction in CHD death alone, a 3% reduction in nonfatal MI alone, and nearly a 4% reduction in stroke. In the VA-HIT study, the effects of HDL-C elevation were seen ≈3 years after study initiation. A recent subanalysis of the VA-HIT study data suggests that the change in HDL-C level was the major factor in prediction of events.20
Although the value of HDL-C as a predictor for cardiac events and outcomes has been observed previously, less is known about the role of HDL-C in the post-CABG patient. Cardiac mortality and total mortality rates were related to serum HDL-C levels (P<0.01) in a post-CABG study of 83 patients who were followed for 10 years. Eighty-nine percent of the patients with an HDL-C level of >38.8 mg/dL survived for 10 years, whereas only 38% of patients with an HDL-C level of <38.8 mg/dL were alive for that same time period.21 We now corroborate this finding in a larger population 15 to 20 years beyond surgery. Previous observations in a 10-year angiographic follow-up study (n=82) found HDL-C to be one of the principal factors in the prediction of new graft disease.22 A several-year delay has been described before the development of these new pathological lesions in grafts in patients after CABG.23 24 This is consistent with the divergence of our mortality curves between high and low HDL-C subgroups at ≈6 years after surgery. Further, new lesion formation in vein grafts has been prevented with gemfibrozil, a benefit directly correlated with increases in serum HDL-C level.25 Thus, HDL-C proves to be an excellent predictive marker for both vein graft progression and death whose modification would potentially prevent the progression of disease in the post-CABG patient.
Our analyses are limited by the nature of the database and the selection. We have identified a population of post-CABG men with a low prevalence of DM and HTN. Inferences to other populations with different risk profiles should be made cautiously. In addition, although men without complete baseline lipid profiles (n=1266) did not differ significantly from the study cohort and although overall survival was similar for these groups, ascertainment biases cannot be totally excluded. Changes in health status, such as the subsequent development of DM or HTN, were not discerned, nor were changes in treatment (eg, antiplatelet drugs, β-blockers, or lipid-lowering therapy). Finally, the exact cause of death is not available in this cohort. These factors therefore cannot be incorporated into the analyses.
In summary, this is the largest study to examine the role of HDL-C in the determination of survival in patients with severe CAD that warrants CABG. We demonstrate that baseline HDL-C is a profound predictor of all-cause mortality and event-free survival beyond other traditional and surgical risk factors. Although the evaluated cohort represents a highly selected group from the late 1970s, these results suggest that men with an HDL-C level of >35 mg/dL are 50% more likely to survive and to avoid recurrent MI or repeat revascularization after CABG. Therefore, during the contemplation of vascular intervention, an HDL-C level should be measured, because it represents a powerful prognostic marker for long-term outcome after CABG. HDL-C may also prove to be a major target for preventive therapies in the post-CABG population.
- Copyright © 2000 by American Heart Association
Ronnemaa T, Laakso M, Kallio V, et al. Serum lipids, lipoproteins, and apolipoproteins and the excessive occurrence of coronary heart disease in non-insulin-dependent diabetic patients. Am J Epidemiol. 1989;130:632–645.
Stamler J, Vaccaro O, Neaton JD, et al. Diabetes, other risk factors, and 12-yr cardiovascular mortality for men screened in the Multiple Risk Factor Intervention Trial. Diabetes Care. 1993;16:434–444.
Bass KM, Newschaffer CJ, Klag MJ, et al. Plasma lipoprotein levels as predictors of cardiovascular death in women. Arch Intern Med. 1993;153:2209–2216.
Assmann G, Schulte H. Relation of high-density lipoprotein cholesterol and triglycerides to incidence of atherosclerotic coronary artery disease (the PROCAM experience). Prospective Cardiovascular Munster study. Am J Cardiol. 1992;70:733–737.
Tuzcu M, Emrie A, Goormastic M, et al. Incidence and prognostic significance of intraventricular conduction abnormalities after coronary bypass surgery. J Am Coll Cardiol. 1990;16:607–610.
Hainline A, Karon J, Lippel K, eds. Manual of Laboratory Operations, Lipid, and Lipoprotein Analysis. Bethesda, MD: Lipid Research Clinics Program, National Heart, Lung, and Blood Institute, US Government Printing Office; 1982; Vol No 1982-361-132:678.
Lauer MS, Blackstone EH, Young JB, et al. Cause of death in clinical research: time for a reassessment? J Am Coll Cardiol. 1999:618–620.
Gottlieb SS. Dead is dead: artificial definitions are no substitute. Lancet. 1997;349:662–663.
Kaplan E, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53:457–481.
Cox D. Regression models and life tables. J R Stat Soc. 1972;B34:187–220.
Efron B. The efficiency of Cox’s likelihood function for censored data. J Am Stat Assoc. 1977;76:312–319.
McNamara JR, Campos H, Ordovas JM, et al. Effect of gender, age, and lipid status on low density lipoprotein subfraction distribution: results from the Framingham Offspring Study. Arteriosclerosis. 1987;7:483–490.
Kannel WB. Metabolic risk factors for coronary heart disease in women: perspective from the Framingham Study. Am Heart J. 1987;114:413–419.
Kannel WB, Wilson PW. An update on coronary risk factors. Med Clin North Am. 1995;79:951–971.
Kannel W, Wilson W. Cardiovascular risk factors and hypertension. In: Izzo J, Black H, eds. Hypertension Primer: The Essentials of High Blood Pressure. Baltimore, Md: Lippincott Williams & Wilkins; 1999:199–202.
Garcia MJ, McNamara PM, Gordon T, et al. Morbidity and mortality in diabetics in the Framingham population: sixteen year follow-up study. Diabetes. 1974;23:105–111.
Kinosian B, Glick H, Garland G. Cholesterol and coronary heart disease: predicting risks by levels and ratios. Ann Intern Med. 1994;121:641–647.
Forde OH, Thelle DS, Miller NE, et al. The Tromso Heart Study: distribution of serum cholesterol between high density and lower density lipoproteins in subjects of Norse, Finnish and Lappish ethnic origin. Acta Med Scand. 1978;203:21–26.
Rubins H, Robins S, Collins D, et al, for the Veterans Affairs High-Density Lipoprotein Cholesterol Intervention Trial Study Group. Gemfibrozil for the secondary prevention of coronary heart disease in men with low levels of high-density lipoprotein cholesterol. N Engl J Med. 1999;341:410–418.
Robins S, Collins D, Rubins H. Relation of baseline lipids: lipid changes with gemfibrozil to cardiovascular endpoints in the VA-High Density Lipoprotein Intervention Trial (VA-HIT). Circulation. 1999;100(suppl I):I-238. Abstract.
Linden T, Bondjers G, Karlsson T, et al. Serum triglycerides and HDL cholesterol–major predictors of long-term survival after coronary surgery. Eur Heart J. 1994;15:747–752.
Campeau L, Enjalbert M, Lesperance J, et al. The relation of risk factors to the development of atherosclerosis in saphenous-vein bypass grafts and the progression of disease in the native circulation: a study 10 years after aortocoronary bypass surgery. N Engl J Med. 1984;311:1329–1332.
Walts AE, Fishbein MC, Matloff JM. Thrombosed, ruptured atheromatous plaques in saphenous vein coronary artery bypass grafts: ten years’ experience. Am Heart J. 1987;114:718–723.
Alderman EL, Corley SD, Fisher LD, et al. Five-year angiographic follow-up of factors associated with progression of coronary artery disease in the Coronary Artery Surgery Study (CASS): CASS Participating Investigators and Staff. J Am Coll Cardiol. 1993;22:1141–1154.
Frick MH, Syvanne M, Nieminen MS, et al. Prevention of the angiographic progression of coronary and vein-graft atherosclerosis by gemfibrozil after coronary bypass surgery in men with low levels of HDL cholesterol: Lipid Coronary Angiography Trial (LOCAT) Study Group. Circulation. 1997;96:2137–2143.