Some Coronary Risk Factors Related to the Insulin Resistance Syndrome and Treatment With Gemfibrozil
Experience From the Helsinki Heart Study
Background Coronary risk factors related to the insulin resistance syndrome tend to cluster in the same individual. Our previous studies have shown that the dyslipidemia characteristic of this syndrome—low HDL cholesterol and high triglyceride (TG) levels—responds well to treatment with gemfibrozil. Most factors related to insulin-resistance syndrome decrease fibrinolytic capacity, whereas a recent study showed that gemfibrozil improves it and thus may attenuate thrombotic events. To discover whether subjects with clustering of factors related to this resistance might in particular benefit from gemfibrozil, we reanalyzed the Helsinki Heart Study data.
Methods and Results We used Cox regression models to explore the effects of gemfibrozil among overweight subjects with additional coronary risk factors in this hypercholesterolemic male population of 2046 subjects randomized to gemfibrozil and 2035 to placebo. The effect of gemfibrozil was largely confined to overweight subjects: among those with body mass index (BMI) >26 kg/m2, the net difference in cardiac end points between gemfibrozil and placebo groups was 21 (25 of 1119 versus 46 of 1081), and in those with BMI ≤26 kg/m2, it was 7 (31 of 927 versus 38 of 954). The risk reduction with gemfibrozil was 78% (P=.002) among those with BMI >26 kg/m2 and dyslipidemia (TG ≥2.3 mmol/L and HDL cholesterol <1.08 mmol/L). Among those with BMI >26 kg/m2 and three or four of the following factors present—smoking, sedentary lifestyle, blood pressure ≥140/90 mm Hg, or blood glucose >4.4 mmol/L—the risk reduction was 68% (P=.03).
Conclusions Gemfibrozil reduced the coronary risk mainly in overweight subjects with additional risk factors known to contribute to the insulin-resistance syndrome or predispose to it.
Whether obesity is an independent risk factor for CHD remains to some degree controversial,1 2 but it is well recognized that obesity tends to entail a clustering of the major CHD risk factors, such as hypertension, high levels of TG, and low levels of HDL-C.3 4 5 A unifying approach to explain this clustering of risk factors has been presented by Reaven,6 with insulin resistance as the primary metabolic disturbance and hyperinsulinemia, hypertriglyceridemia, low level of HDL-C, and hypertension as secondary features. Lifestyle factors such as smoking and low physical activity may also enhance the progression of the insulin-resistance syndrome.7 8 In addition, there is much evidence supporting the view that this syndrome or its components simultaneously exert an unfavorable effect on the blood coagulation and fibrinolytic systems, which thereby predisposes to thrombosis9 10 11 and increased CHD risk.12 13
The Helsinki Heart Study was a primary prevention trial to test the hypothesis that gemfibrozil reduces CHD incidence in middle-aged dyslipidemic men.14 During the 5-year trial, a 34% reduction of cardiac end points was seen,15 and later analyses showed that the greatest benefit was derived by those with an elevated ratio of LDL-C to HDL-C and a high level of TG.16 The protective effect was estimated to have been achieved via modulation of the lipid levels. However, recent studies have shown that in addition to its effect on lipids, gemfibrozil may modulate the fibrinolytic system.17 18
These findings suggest that in CHD prevention, even subjects with some nonlipid risk factors, especially those related to the insulin resistance syndrome or predisposing to it, may in particular benefit from treatment with gemfibrozil. Forming different subgroups of obesity, smoking, hypertension, sedentary lifestyle, elevated blood glucose and levels of HDL-C and TG, and their combinations allowed us to delineate subgroups with higher and lower probabilities of the insulin resistance syndrome and to compare the treatment effect in these subgroups.
The Helsinki Heart Study was a 5-year, double-blind clinical trial to test the hypothesis that lowering serum LDL-C and TG levels and elevating serum HDL-C level with gemfibrozil would reduce cardiac end points in dyslipidemic men. The design and principal results of the trial have been described earlier.14 15 Briefly, the participants for the study were selected from 23 531 men aged 40 to 55 years who were employed by two state agencies and five industrial companies in Finland. Volunteers were selected for the trial by two successive screenings with the criteria that their non–HDL-C should be ≥5.2 mmol/L and they should have no evidence of CHD or other major disease. The trial participants were randomly allocated to gemfibrozil (n=2046) or placebo (n=2035). Definite fatal and nonfatal myocardial infarctions and cardiac deaths were the end points recorded.14
Measurement of Lipids, Lipoproteins, and Glucose
The samples from the local clinics were mailed daily to the central laboratory (at the National Public Health Institute in Helsinki). The interval from sampling to analysis ranged from 1 to 5 days. Total cholesterol concentration was determined directly from serum, and HDL-C was measured after precipitation of VLDL and LDL with dextran sulfate/magnesium chloride by an enzymatic method (Kit No. 236691, Boehringer Mannheim). The serum concentration of TG was measured as glycerol after enzymatic hydrolysis with lipase/esterase (Kit No. 124966, Boehringer Mannheim). The LDL-C concentration was calculated with the formula LDL-C=total cholesterol minus HDL-C minus TG divided by 2.2. TG levels of ≥8.1 mmol/L (≥717 mg/dL) were excluded from LDL calculations, as were computed LDL-C levels of <2.6 mmol/L (<101 mg/dL). These restrictions resulted in 1.4% missing values for LDL-C. Fasting samples for determining serum TG were drawn at the second screening visit and thereafter semiannually.
The lipid and lipoprotein values were measured and calculated similarly for each follow-up visit and the mean of all available values during the follow-up represents the subjects’ in-trial value. In some analyses, dichotomized variables were used. For TG, we used 2.3 mmol/L (204 mg/dL) as the cutoff point according to European recommendations for treatment practice.19 As no such recommendations existed for HDL-C, we used the limit value for the lowest tertile in the trial population (1.08 mmol/L [42 mg/dL]). Blood glucose was determined from fasting samples at the second screening visit and annually during the trial. The mean value in the trial population, 4.4 mmol/L (79.3 mg/dL), was used as the dichotomy cutoff point (median, 4.3 mmol/L).
Recording and Categorization of Lifestyle Factors and Blood Pressure
Obesity was measured in terms of BMI (in kg/m2). In most analyses, it was dichotomized with 26 kg/m2 as the cutoff point (median, 26.3 kg/m2; mean±SD, 26.6±2.9 kg/m2). In some analyses, the group with BMI >26 kg/m2 was further divided into “overweight” and “obese” with the cutoff point at 30 kg/m2, the commonly used limit for actual obesity. Smoking habits were classified at baseline based on the reported number of cigarettes smoked per day. For the analyses, the subjects were classified as nonsmokers or smokers. All ex-smokers who had stopped more than 3 months before the beginning of the study were categorized as nonsmokers. Spare time physical activity was recorded according to the Gothenburg scale20 into four categories, but for the analyses we used a dichotomized scale: sedentary (categories I and II) and active (categories III and IV).
The baseline measurements of systolic and diastolic blood pressures were used to scale hypertension: 1, systolic <140 mm Hg and/or diastolic <90 mm Hg; 2, systolic ≥140 mm Hg and diastolic ≥90 mm Hg but not level 3; and 3, systolic ≥165 mm Hg and diastolic ≥95 mm Hg.
We present the mean values of HDL-C, LDL-C, and TG at baseline and during treatment by categories of the other risk factors to explore the effect of treatment on lipid levels. Strictly, the effect of treatment in any subgroup is the difference between placebo and gemfibrozil mean values during treatment. However, as the placebo and gemfibrozil mean values at baseline were both closely similar to the placebo mean during the trial, only the mean values for the gemfibrozil group are presented. The P values for the differences in mean lipid levels were derived with a two-way ANOVA. In these calculations, the log of TG was used. The crude incidences of CHD in different subgroups were given to illustrate the differences in treatment effect. In some instances, the results were confirmed by calculating the relative risks using Cox’s proportional hazards models with appropriate covariates.21 The treatment effect was expressed in terms of reduced relative risk in the gemfibrozil subgroup compared with the corresponding placebo group. The P value for this comparison of relative risks was based on likelihood ratio statistics and was obtained via Cox models.
Mean Lipid Levels
Both at baseline and during the trial, overweight subjects (BMI >26 kg/m2) in the gemfibrozil group had lower HDL-C (P<.001) and higher TG (P<.001) levels than lean subjects (BMI <26 kg/m2) (Table 1⇓). The levels of TG rose with increasing blood pressure (baseline, P=.005; trial, P=.007), but the effect was less pronounced than with BMI. Treatment with gemfibrozil induced a significant decrease in levels of LDL-C and TG and an increase in HDL-C in all categories of blood pressure, BMI, and fasting blood glucose. Both smoking and sedentary lifestyle were accompanied by a lower baseline level of HDL-C and higher level of TG (Table 2⇓). The impact of smoking on TG level (P<.001) was somewhat greater than that of physical activity (P=.13).
The prevalence of subjects with both high TG and low HDL-C levels in the total study population was 14%. Among those with hypertension, elevated blood glucose, sedentary lifestyle, or smoking, the prevalence was higher (Table 3⇓), although body mass had a greater impact than these other variables.
Overweight as a Coronary Risk Factor
The treatment effect was mainly confined to overweight subjects. The total difference in cardiac end points between the placebo and the gemfibrozil groups was 28, with 21 end points among those with BMI >26 kg/m2 (25 of 1119 versus 46 of 1081) and only 7 among those with BMI <26 kg/m2 (31 of 927 versus 38 of 954) (Fig 1⇓). The risk pattern was slightly J shaped with 66% excess risk among those with BMI >30 kg/m2 compared with those with BMI ≤26 kg/m2 (Table 4⇓). The risk estimate was strongly dependent on the covariates included in the model: adding age and smoking increased the risk estimate for BMI (from 66% to 78%), whereas adding blood pressure and HDL-C considerably decreased (to 37%) the estimate of CHD risk ascribed to obesity.
Joint Effect of Overweight and Low HDL-C and High TG Levels
Among those with BMI >26 kg/m2, the subgroup with high TG and low HDL-C levels showed the greatest CHD risk: the CHD incidence was 2.6 times that found in the subgroup with normal TG and normal HDL-C levels (Fig 2⇓). However, the high-risk group experienced a substantial treatment effect, with a 78% reduction in CHD incidence (P=.002).
Joint Effect of Overweight and Other CHD Risk Factors
Obesity accentuated the relative risks associated with sedentary lifestyle, hypertension, and blood glucose, whereas the relative risk associated with smoking was the same among lean and overweight men (Fig 3⇓). However, the treatment effect was consistently most favorable in the overweight subjects. Among the 30% of obese subjects who had at least three of these four risk factors present, 19 cardiac end points were “prevented,” whereas only 2 were prevented among the remaining 70% of obese subjects (Fig 4⇓, top). In terms of relative risk, those of the placebo subjects with BMI >26 kg/m2 and at least three of the other risk factors present had a relative risk of 5.3 (95% confidence interval, 2.2 to 12.9) compared with those with BMI >26 kg/m2 and none or one other risk factor, whereas the high-risk subjects receiving gemfibrozil experienced a risk reduction of 68% compared with high-risk placebo-treated men. The placebo arm of the high-risk subgroup with high TG and low HDL-C levels had a very high CHD incidence: 32.7 (per 1000 person-years), with 12 cardiac end points (Fig 4⇓, bottom). There also was a good treatment effect; the incidence in the corresponding gemfibrozil subgroup was 5.5, with only 2 end points. However, the subgroup with normal TG and normal HDL levels also profited from treatment, showing a 57% reduction of cardiac end points.
Our earlier report on the Helsinki Heart Study showed that treatment with gemfibrozil considerably reduced the CHD risk in dyslipidemic subjects with high TG and low HDL-C levels.16 In the present article, we extended the concept and showed that obesity greatly enhances the CHD risk found in this subgroup but also that a considerable treatment benefit occurs in this combination of obesity and dyslipidemia. Similarly, we found that although obesity enhanced the CHD risk due to nonlipid risk factors such as smoking, hypertension, sedentary lifestyle, and elevated blood glucose, substantial treatment effects were also seen in these subgroups.
The Lean and the Overweight: CHD Risk and Effect of Treatment
In the present study, obesity was measured in terms of BMI, which is not the most accurate way of determining abdominal obesity, the risk factor for CHD.4 Moreover, when the cutoff point for obesity was set at 26 kg/m2, the CHD risk was diluted because the risk due to obesity increased notably only when BMI was >30 kg/m2. Nevertheless, there was a considerable treatment effect of 43% even among subjects merely with overweight (BMI, 26 to 30 kg/m2), and this effect persisted even when other risk factors were adjusted for (Table 4⇑). The treatment effects, as well as the risks, mirror the underlying metabolic differences among the lean and the obese. Among the overweight, those with elevated blood glucose, hypertension, or sedentary lifestyle or who were smokers consistently had a substantial treatment effect, whereas lean subjects with similar attributes experienced no treatment effect (Fig 3⇑). The risk pattern among lean and obese hypertensives in the placebo group was unexpected, however, as several studies have found that lean hypertensives are at greater risk of CHD than obese hypertensives.22 23 Carman et al23 even suggested that hypertension in lean individuals may be a distinct disease with strong genetic determinants. The marked discrepancy of our findings from those of other studies was possibly caused by the selection criterion for our subjects, which required them to be hyperlipidemic. This could entail an enhanced role for hypofibrinolysis among the obese hypertensives with an accompanying increase in CHD risk.24 In any case, our findings suggest that different metabolic mechanisms regulate the treatment effect among the lean and the obese.
Distribution of Insulin Resistance in Study Population
Several studies have shown that obesity is related to insulin resistance, and a recent Finnish study showed that the presence of type IIB hyperlipidemia increased insulin resistance related to obesity alone.25 Due to the selection criterion, the majority of the Helsinki Heart Study population had high levels of LDL-C and a higher proportion of type IIB subjects than the general population. The prevalence of insulin resistance among the obese may thus be higher in our study population than in the general population.
Lamarche et al26 found that the groups with the lipid level combinations of high TG–low HDL-C, high TG–normal HDL-C, and normal TG–low HDL-C were metabolically different. Only the group with high TG–low HDL-C had higher fasting plasma insulin levels and hyperinsulinemia during an oral glucose tolerance test compared with normolipemic subjects, which was suggestive of insulin resistance. In our study, higher levels of obesity involved greater proportions of subjects with high TG and low HDL-C levels, and thus they very probably also had higher prevalences of insulin resistance. Our results also indicated that lifestyle factors such as smoking and little spare-time physical activity were associated with a higher prevalence of high TG and low HDL-C levels. Previous studies have shown that both smoking and sedentary lifestyle may promote the insulin-resistance syndrome.7 8
Viewed in the context of findings from other studies, results from the present study suggest that the metabolic state characterized by insulin resistance may be a common feature underlying the treatment effects seen. However, these inferences remain mainly on a hypothetical level: first, because insulin resistance was not measured in our study, and second, because all of the risk factors considered may occur without insulin resistance and insulin resistance may occur without them, although far less frequently.
Some Possible Pathways for the Treatment Effect
There are several putative pathways for the treatment effect seen. First, the good treatment effect among those with the nonlipid risk factors related to the insulin-resistance syndrome could be caused by a direct effect of gemfibrozil on insulin resistance. However, as far as we know, there are no studies reporting such an effect.
Second, a high level of TG and a low level of HDL-C, the dyslipidemia characteristic of insulin-resistance syndrome, responds well to treatment with gemfibrozil16 (Tables 1⇑ and 2⇑). Therefore, part of the treatment effect seen in subjects with nonlipid risk factors may certainly be ascribed to a correction of the accompanying dyslipidemia. Table 3⇑ shows that the prevalence of this kind of dyslipidemia did increase with rising levels of any of the nonlipid risk factors, whereas LDL-C remained fairly constant across categories of the nonlipid risk factors (Tables 1⇑ and 2⇑).
Third, there is ample evidence indicating that TG influences hemostatic functions, especially the level of plasminogen activator inhibitor–1.27 The particular pattern of dyslipidemia in the subgroups considered suggests that along with a lowering of the TG level, gemfibrozil may have enhanced fibrinolysis. In the Helsinki Heart Study, there were fewer fatal cerebral infarctions in the gemfibrozil group during the trial (one versus four) and more fatal intracranial hemorrhages (five versus none) than in the placebo group, and this trend was accentuated when the entire extended 8.5-year follow-up was considered (cerebral infarctions, 1 versus 5; intracranial hemorrhages, 7 versus 1).28
Fourth, as suggested by Fujii and Sobel,18 gemfibrozil may also potentiate fibrinolysis by direct diminution of endogenous plasminogen activator inhibitor–1 synthesis; they even proposed that the treatment effect in the Helsinki Heart Study may have been at least in part related to such a diminution.
In addition to the favorable modification of the lipid and lipoprotein cholesterol levels, there are several other putative pathways for the treatment effects seen. When considering our findings with the experimental and epidemiological evidence from other studies,18 27 29 the fibrinolytic pathway—either via TG or direct—seems plausible. Such conclusions remain speculative, however, as no fibrinolytic parameters were measured in the Helsinki Heart Study.
Some Aspects Related to the Epidemiological Modeling
The assessment of CHD risk due to obesity (Table 4⇑) illustrates the difficulties of measuring effect when the risk factors form a web of interdependencies. We are exposed to these difficulties even when using the regression models designed for risk assessment. For example, in Table 4⇑, when age and smoking were added to the model, the fit improved and the risk due to BMI increased. This probably occurred because smoking explained the end points found among the lean subjects and thereby decreased the variance around the BMI risk estimate. The addition of systolic blood pressure and HDL-C to the model had the opposite effect; the risk estimate for BMI decreased. This occurred because BMI is associated with both systolic blood pressure and HDL-C, and part of its atherogenic effect is probably mediated through systolic blood pressure and low HDL-C (and simultaneous high TG). When adjusting for HDL-C, we thus adjusted for part of the effect of BMI on CHD. There are similar problems with the other risk factors considered here: they are interrelated, they progress to a large extent simultaneously, and they probably in part share the same pathways in their atherogenic effect. In these circumstances, the notion of “independent effect” has no great relevance.
In conclusion, gemfibrozil reduced the coronary risk mainly in overweight subjects who had one or more additional risk factors known to contribute to the insulin-resistance syndrome or predispose to it. With increased clustering of hypertension, elevated blood glucose level, smoking, and sedentary lifestyle among the overweight, the CHD risk rose, as expected, but the treatment effect also improved.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|CHD||=||coronary heart disease|
- Received February 23, 1995.
- Revision received April 12, 1995.
- Accepted May 3, 1995.
- Copyright © 1995 by American Heart Association
Barrett-Connor EL. Obesity, atherosclerosis, and coronary heart disease. Ann Intern Med. 1985;103:1010-1019.
Stanton JL, Braitman LE, Riley AM, Khoo C-S, Smith JL. Demographic, dietary, life style, and anthropometric correlates of blood pressure. Hypertension. 1982;4(suppl III):III-135-III-142.
Després JP, Moorjani S, Lupien PJ, Tremblay A, Nadeau A, Bouchard C. Regional distribution of body fat, plasma lipoproteins, and cardiovascular disease. Arteriosclerosis. 1990;10:497-511.
Pouliot M-C, Després J-P, Nadeau A, Moorjani S, Prud’Homme D, Lupien PJ, Tremblay A, Bouchard C. Visceral obesity in men: associations with glucose, plasma insulin, and lipoprotein levels. Diabetes. 1992;41:826-834.
Reaven GM. Role of insulin resistance in human disease. Diabetes. 1988;37:1595-1607.
Juhan-Vague I, Vague P, Alessi MC, Badier C, Valadier Aillaud MF, Atlan C. Relationships between plasma insulin, triglyceride, body mass index, and plasminogen activator inhibitor 1. Diabetes Metab. 1987;13:331-336.
Hoffman CJ, Miller RH, Hultin MB. Correlation of factor VII activity and antigen with cholesterol and triglycerides in healthy young adults. Arterioscler Thromb. 1992;12:267-270.
Nordt TK, Schneider DJ, Sobel BE. Augmentation of the synthesis of plasminogen activator inhibitor type-1 by precursors of insulin: a potential risk factor for vascular disease. Circulation. 1994;89:321-330.
Mänttäri M, Elo O, Frick MH, Haapa K, Heinonen OP, Heinsalmi P, Helo P, Huttunen JK, Kaitaniemi P, Koskinen P, Manninen V, Mäenpää H, Mälkönen M, Norola S, Pasternack A, Pikkarainen J, Romo M, Sjöblom T, Nikkilä EA. The Helsinki Heart Study: basic design and randomization procedure. Eur Heart J. 1987;8:1-29.
Frick MH, Elo O, Haapa K, Heinonen OP, Heinsalmi P, Helo P, Huttunen JK, Kaitaniemi P, Koskinen P, Manninen V, Mäenpää H, Mälkönen M, Mänttäri M, Norola S, Pasternack A, Pikkarainen J, Romo M, Sjöblom T, Nikkilä EA. The Helsinki Heart Study: primary prevention trial with gemfibrozil in middle-aged men with dyslipidemia: safety of treatment, changes in risk factors, and incidence of coronary heart disease. N Engl J Med. 1987;317:1237-1245.
Manninen V, Tenkanen L, Koskinen P, Huttunen JK, Mänttäri M, Heinonen OP, Frick MH. Joint effects of serum triglyceride and LDL cholesterol and HDL cholesterol concentrations of coronary heart disease risk in the Helsinki Heart Study: implications for treatment. Circulation. 1992;85:37-45.
Fujii S, Sobel BE. Direct effects of gemfibrozil on the fibrinolytic system: diminution of synthesis of plasminogen activator inhibitor type 1. Circulation. 1992;85:1888-1893.
Study Group, European Atherosclerosis Society. The recognition and management of hyperlipidaemia in adults: a policy statement of European Atherosclerosis Society. Eur Heart J. 1988;9:571-600.
Wilhelmsen L, Tibblin G, Fodor J, Werkö L. A multifactorial primary preventive trial in Gothenburg, Sweden. In: Larson OA, Malmborg RO, eds. Coronary Heart Disease and Physical Fitness. Copenhagen: Munksgaard; 1971:266-270.
Dixon WJ, ed. BMDP Statistical Software Manual. Berkeley/Los Angeles/Oxford: University of California Press; 1990.
Cambien F, Chretien JM, Ducimetiere P, Guize L, Richard JL. Is the relationship between blood pressure and cardiovascular risk dependent on body mass index? Am J Epidemiol. 1985;122:434-442.
Carman WJ, Barrett-Connor E, Sowers MF, Khaw K. Higher risk of cardiovascular mortality among lean hypertensive individuals in Tecumseh, Michigan. Circulation. 1994;89:703-711.
Karhapää P, Voutilainen E, Malkki M, Laakso M. Obese men with type IIB hyperlipidemia are insulin resistant. Arterioscler Thromb. 1993;13:1469-1475.
Lamarche B, Després J-P, Pouliot M-C, Prud’homme D, Moorjani S, Lupien PJ, Nadeau A, Tremblay A, Bouchard C. Metabolic heterogeneity associated with high plasma triglyceride or low HDL cholesterol levels in men. Arterioscler Thromb. 1993;13:33-40.
Mussoni L, Mannucci L, Sirtori M, Camera M, Maderna P, Sironi L, Tremoli E. Hypertriglyceridemia and regulation of fibrinolytic activity. Arterioscler Thromb. 1992;12:19-29.