Gemfibrozil Treatment of Combined Hyperlipoproteinemia
No Improvement of Fibrinolysis Despite Marked Reduction of Plasma Triglyceride Levels
Abstract Hypertriglyceridemia is linked to impaired fibrinolytic function, and lipid-lowering treatment with fibric acid derivatives could hypothetically improve fibrinolysis in this condition. We therefore conducted a double-blind, placebo-controlled, crossover study of gemfibrozil treatment on fibrinolytic function in 21 men with combined hyperlipoproteinemia. Measurements were performed at rest and during mental stress and after venous occlusion. The patients had clearly disturbed fibrinolytic function, with elevated plasminogen activator inhibitor–1 (PAI-1) activity at rest (≈25 U/mL; reference, <15 U/mL). Gemfibrozil reduced plasma total and VLDL cholesterol as well as all triglyceride fractions, whereas HDL cholesterol increased (P<.001 for all). Total triglyceride levels were reduced by 57±4% (from 5.3 to 2.1 mmol/L). Fasting serum insulin levels were not altered by gemfibrozil treatment. Plasma levels of PAI-1 activity and tissue-type plasminogen activator (TPA) activity or antigen were unaffected by gemfibrozil treatment both at rest and during the provocations. The levels of d-dimer, plasmin/antiplasmin complex, and fibrinogen were also uninfluenced by gemfibrozil treatment. Mental stress elevated plasma TPA (P=.0036) and lowered PAI-1 (P=.0012) activity during placebo but not gemfibrozil treatment (P=.28 and P=.17, respectively), but treatment effects did not differ by ANOVA on Δ values (ie, stress minus rest). Venous occlusion reduced PAI-1 activity, whereas TPA and plasmin/antiplasmin complex increased during both treatments. Thus, gemfibrozil treatment did not improve fibrinolysis or lower fibrinogen levels in men with combined hyperlipoproteinemia despite marked reduction of plasma triglyceride levels. It seems unlikely that improved fibrinolysis explains the primary preventive effect of gemfibrozil.
- plasminogen activator inhibitor
- tissue-type plasminogen activator
- Received May 28, 1995.
- Accepted November 2, 1995.
Impaired fibrinolytic function may predispose susceptible individuals to thrombosis and the development of coronary artery disease,1 and hypertriglyceridemia is associated with a global impairment of fibrinolytic function2 3 and an elevation of PAI-1.4 Thus, deranged fibrinolysis may be an important factor in patients with hypertriglyceridemia that could lead to an increased risk of coronary artery disease.
The primary preventive Helsinki Heart Study showed that lipid-lowering treatment with gemfibrozil, which markedly lowers plasma triglyceride levels, reduces the incidence of coronary heart disease in hyperlipidemic men.5 Retardation of lipid-dependent coronary atherosclerosis is one possible explanation for the favorable results, but beneficial effects on fibrinolysis may have contributed. Indeed, high concentrations of gemfibrozil decrease the release of PAI-1 from vascular endothelial cells in vitro.6 Furthermore, when given to patients with hypertriglyceridemia or survivors of myocardial infarction, gemfibrozil has been claimed to improve certain fibrinolytic parameters.7 8 9 10 11 However, these findings should be interpreted cautiously since two of the trials were uncontrolled,7 8 and another two relied on subgroup analysis to demonstrate effects.10 11 Moreover, the studies were limited in size.
The fibrinolytic system is usually studied at rest, but standardized provocations, such as mental stress or venous occlusion, may be used to obtain information on the dynamic regulation of the system. In other words, a fibrinolytic system that performs well at rest may not respond properly during stressful situations.
The present study was performed to clarify whether gemfibrozil alters the fibrinolytic function in men with combined hyperlipoproteinemia. A double-blind, placebo-controlled, crossover design was used, and measurements were performed both at rest and during provocation by mental stress and venous occlusion.
Patients and Study Protocol
Twenty-one men aged 21 to 70 (mean, 47) years with total plasma cholesterol concentrations ≥6.4 mmol/L and triglyceride levels ≥2.3 mmol/L participated in the study. Patients with secondary hyperlipoproteinemia due to hypothyroidism, diabetes mellitus, and nephrosis were excluded, as were those who had experienced an acute myocardial infarction or unstable angina pectoris during the preceding 12 months. Two patients received concomitant anti-ischemic medication with β-blockers and a calcium antagonist. Another 3 patients had hypertension, for which they received diuretics, β-blockers, and an angiotensin-converting enzyme inhibitor. In addition, 1 hypertensive patient was treated with allopurinol, and another patient received clomipramine. All nonstudy medication was kept constant during the study. Body weight and BMI of the patients at randomization were 87±3 kg and 27.7±0.8 kg/m2 (mean±SEM), respectively. Six patients were smokers.
The study design is shown in panel A of the Figure⇓. The trial started with a dietary period of 4 to 6 weeks, during which the patients were instructed to follow the AHA step I diet. Previous lipid-lowering medication (six patients) was discontinued at the onset of the diet period. After the dietary run-in phase the patients were randomized to receive gemfibrozil 600 mg BID (Parke-Davis) or placebo in a double-blind fashion. After 10 to 12 weeks of treatment the patients were switched to the alternate treatment.
The design of individual experiments is shown in panel B. The experiments were performed at the end of the treatment periods between 8:30 and 11:30 am, while the subjects were in the fasting state. Venous blood was collected via 19G butterfly needles or evacuated tube systems (after venous occlusion) into trisodium citrate (Becton-Dickinson) or citric acid/citrate (Stabilyte, Biopool AB). Samples were obtained after a 1-hour rest, after 15 minutes of mental stress (a modified Stroop Color Word Conflict Test12 ), and again after 10 minutes of venous occlusion. The venous occlusion test was performed by inflating a BP cuff on the upper arm to 100 mm Hg for 10 minutes. Blood was then collected before the cuff was deflated. The samples were centrifuged at 4°C for 10 minutes at 1500g, and the plasma was divided into aliquots and stored frozen at −80°C until analysis.
Compliance was monitored by tablet count. The manufacturer randomized the treatment order (11 started on placebo), and the participants were instructed to abstain from smoking and caffeine-containing beverages on the experimental days.
The study was approved by the ethics committee of the Karolinska Institute. All patients gave their informed consent to participate.
Assessments of Fibrinolytic Variables and Fibrinogen
PAI-1 activity and TPA antigen were determined on citrated plasma samples by using commercially available kits (Spectrolyze PL and TintElize tPA, respectively, Biopool AB). Determinations of TPA activity (Spectrolyze fibrin, Biopool), fibrin degradation product d-dimer (TintElize d-dimer, Biopool), and PAP were performed on acidified plasma samples (Stabilyte, Biopool). Fibrin d-dimer and PAP assays provide more dependable results in acidified plasma samples than in normal citrated samples (B. Wiman, M. Haegerstrand-Björkman, unpublished data, 1994). Because significantly higher concentrations of these compounds were found in citrated than in acidified samples, suggesting in vitro generation during the handling of citrated samples, acidified plasma samples were chosen for assays of PAP and fibrin d-dimer.
The method for determination of PAP in plasma was a double-antibody enzyme-linked immunosorbent assay developed in our laboratory. The acidified plasma samples were diluted 10-fold with 0.1 mol/L sodium phosphate buffer and passed through small columns of lysine-Sepharose. The adsorbed PAP and plasminogen were eluted with 0.1 mol/L 6-aminohexanoic acid in the phosphate buffer. Subsequently PAP was determined by enzyme-linked immunosorbent assay with goat antiplasmin IgG as catch antibodies and horseradish peroxidase–conjugated antiplasminogen IgG for detection. Reference values are 0.90±0.27 mg/L (mean±SD). Intra-assay and interassay coefficients of variation were <10% at PAP concentrations of ≈1 mg/L. The detection limit was ≈0.02 mg/L. Accuracy, determined as analytical recovery of added pure PAP, was >90% at a concentration of 1 mg/L.
Fibrinogen was determined as a modified thrombin time.13 Fasting serum insulin levels were determined by radioimmunoassay (Insulin RIA 100, Pharmacia).
Lipid and Lipoprotein Determinations
Plasma lipoproteins were quantified by using a combination of preparative ultracentrifugation and precipitation of apoB-containing lipoproteins with phosphotungstic acid.14 15 The cholesterol and triglyceride contents of the various lipoprotein fractions were assessed by using standard enzymatic techniques (Boehringer-Mannheim). ApoA-I and apoB were analyzed by using immunoturbidimetric methods (Orion Diagnostica), and apo(a) levels were determined by radioimmunoassay (Pharmacia Diagnostics).
The catecholamine concentrations in venous plasma were determined by using cation-exchange high-performance liquid chromatography with amperometric detection.16 BPs and heart rates were monitored with an Ohmeda 2300 FINAPRESS BP monitor (Ohmeda Monitoring Systems).
Data are presented as mean±SEM or as median with 25th and 75th percentiles when skewed. Differences between the treatments were evaluated by using a three-way ANOVA according to the standard 2×2 crossover model, controlling for subject and period effects. Skewed data were log-transformed before this evaluation. Changes induced by mental stress and venous occlusion during each treatment were analyzed by using Wilcoxon’s signed rank test followed by Bonferroni corrections for multiple hypothesis testing. Differences between the treatments for the provocation-induced changes were analyzed by using Δ values (ie, stress minus rest). The subgroup analysis of obese (BMI≥25 kg/m2) versus nonobese (BMI<25 kg/m2) patients was performed by using an unpaired Student’s t test or a Mann-Whitney U test. Tests to examine possible carryover effects were applied according to standard procedures,17 and correlations were tested by calculation of Spearman’s rank correlation coefficient. Statistical evaluations were performed by using SuperANOVA and StatView 4.0 software (Abacus Concepts Inc).
Two of the 21 patients did not complete the study, one because of noncompliance and the other because of an acute myocardial infarction. Compliance with medication treatment was good (94±1% of the tablets were consumed during placebo and 93±2% during gemfibrozil treatment). No significant carryover effects between treatment periods were noted for total triglyceride or PAI-1 levels.
Lipid, Lipoprotein, and Insulin Levels
Gemfibrozil treatment significantly lowered all plasma triglyceride fractions as well as total and VLDL cholesterol (Table 1⇓). Total plasma triglyceride concentrations were reduced by 57±4%, whereas HDL cholesterol increased by 22±5%. LDL cholesterol, apoA-I, apoB, and apo(a) levels were unaltered. Fasting serum insulin levels were 14.9±2.5 mU/mL during placebo and 13.2±2.2 mU/mL during gemfibrozil therapy (P=.26).
No significant difference between the treatments was observed for the levels of fibrinolytic variables or fibrinogen either at rest or during mental stress or after venous occlusion (Table 2⇓). Mental stress elevated plasma catecholamine levels, heart rate, and BPs similarly during the two treatments (Table 3⇓). In addition, mental stress elevated plasma TPA activity (P=.0036) and lowered PAI-1 activity (P=.0012) during placebo but not during gemfibrozil (P=.28 and P=.17, respectively) treatment. However, treatment effects did not differ by ANOVA (P=.14 and P=.10 between treatments for Δ values; Table 2⇓). The levels of TPA antigen, d-dimer, PAP, and fibrinogen were not altered by mental stress. Venous occlusion reduced PAI-1 activity more clearly during placebo treatment (placebo, P=.0012; gemfibrozil, P=.12), whereas TPA activity, TPA antigen, and PAP increased significantly during both treatments. d-dimer levels were unaltered during venous occlusion.
Subgroup Analysis of Obese Versus Nonobese Patients
A subgroup analysis was undertaken to compare obese and nonobese (BMI, 29.4±0.7 and 23.6±0.6 kg/m2, respectively) patients. Overweight was associated with higher insulin (16.6±2.4 versus 7.0±1.1 mU/mL, P=.037) and increased total triglyceride (6.4±0.9 versus 3.2±0.3 mmol/L, P=.023) levels. Furthermore, the obese individuals had higher PAI-1 and lower TPA activity levels during mental stress (Table 4⇓). Gemfibrozil treatment did not, however, alter the activities of PAI-1 or TPA in either obese or nonobese patients (Table 4⇓). Individual data showed that gemfibrozil reduced PAI-1 activity in 7 of 12 obese patients at rest, in 6 of 12 during mental stress, and in 3 of 10 after venous occlusion. Corresponding data for TPA levels were elevations in 7 of 12 at rest, 6 of 12 during mental stress, and 7 of 11 after venous occlusion. Thus, there were no significant effects in obese patients.
PAI-1 activity at rest was positively correlated with total triglyceride (placebo, r=.45 and P=.044; gemfibrozil, r=.52 and P=.027), VLDL triglyceride (placebo, r=.45 and P=.046; gemfibrozil, r=.56 and P=.019), and fasting insulin (placebo, r=.78 and P=.001; gemfibrozil, r=.76 and P=.0033) levels. In addition, TPA antigen was positively correlated with total triglyceride (placebo, r=.52 and P=.019; gemfibrozil, r=.48 and P=.041) and VLDL triglyceride (placebo, r=.52 and P=.021; gemfibrozil, r=.50 and P=.034) levels. The gemfibrozil-induced reductions of total and VLDL triglyceride levels were positively correlated with changes in TPA antigen at rest (r=.65, P=.0056 and r=.67, P=.0046, respectively).
The results of this double-blind, placebo-controlled trial do not support the view that gemfibrozil treatment improves fibrinolysis in men with combined hyperlipoproteinemia, despite marked reductions of plasma triglycerides and clear-cut signs of disturbed fibrinolysis. Thus, a causal and direct link between plasma triglyceride levels and fibrinolysis is unlikely in this group of patients.
Case-control studies have shown that patients suffering from coronary artery disease exhibit elevated plasma PAI-1 activity18 19 and that increased PAI-1 levels are related to reinfarction in young survivors of myocardial infarction.20 Moreover, the PLAT study has shown that high PAI-1 levels increase the risk of experiencing thrombotic ischemic events in atherosclerotic patients.21
The reports that gemfibrozil treatment reduces plasma PAI-1 levels in patients with isolated hypertriglyceridemia both at rest7 8 9 and during venous occlusion9 are therefore interesting. Our results did not confirm this favorable effect of gemfibrozil on plasma PAI-1 levels in patients with combined hyperlipoproteinemia, either at rest or after mental stress or venous occlusion. If anything, the provocations lowered PAI-1 activity more consistently during placebo treatment than during active treatment, and the TPA increase after mental stress was significant only during the placebo phase. The discrepancy between the trials could be related to differing study populations or methodological differences. A noteworthy difference between the studies is that basal PAI-1 activity was in the normal range in the previous studies,7 8 9 whereas PAI-1 levels in our patient population were well above the upper reference limit (<15 U/mL). Our PAI-1 data agree with those of others on gemfibrozil-treated hypertriglyceridemic patients with histories of venous thrombosis22 or myocardial infarction.10
The idea that fibric acid derivatives could improve fibrinolysis originates from the fact that these drugs markedly lower plasma triglyceride levels and that a fairly strong positive correlation exists (also present in this study) between triglyceride and PAI-1 levels.18 23 24 It has been claimed that a reduction of plasma triglycerides to levels <2.8 mmol/L is necessary to reduce PAI-1 levels.11 Sixteen of the patients in the present trial met this criterion. Even though total triglyceride levels were reduced by almost 60% in these patients there was no consistent change in PAI-1 activity or any other fibrinolytic variable. In the entire study population gemfibrozil-induced reductions of total and VLDL triglyceride levels correlated positively with changes in TPA antigen at rest, but 8 patients increased their TPA levels during gemfibrozil therapy despite triglyceride reductions. A closer look at these individuals revealed no common features that could explain the results.
It has been convincingly shown that high plasma insulin levels and/or BMI are associated with elevated PAI-1 levels.23 Moreover, reductions of these variables are accompanied by reductions of plasma PAI-1.25 26 Thus, the links to insulin seem to be of major importance, and Juhan-Vague et al27 have suggested that insulin resistance is the key factor regulating plasma PAI-1 levels. As insulin levels are positively related to BMI,28 weight reduction is a plausible explanation for the beneficial effects noted during dietary reduction of triglycerides.2 29 30 Body weight was, unfortunately, not monitored throughout the present investigation, but the dietary background was kept constant, and there are no data showing that gemfibrozil alters body weight profoundly. Our subgroup analysis of obese versus nonobese patients confirmed findings showing that obesity is associated with elevated PAI-1 and insulin levels.25 Gemfibrozil treatment did not influence either insulin or PAI-1 levels or TPA activity in either subgroup, but the sample size is small, and firm conclusions cannot be drawn. Nonetheless, it is quite clear from the present and previous data that there is no simple link between triglyceride levels and PAI-1 activity.
Avellone et al7 report that euglobulin clot lysis time is shortened and fibrinogen levels are reduced in hypertriglyceridemic patients on gemfibrozil therapy. When evaluating these data, it should be borne in mind that the euglobulin clot lysis time method is sensitive to alterations in fibrinogen levels. In the postinfarction trial of Andersen et al10 fibrinogen levels and euglobulin clot lysis time measurements were unaltered, but a subgroup of patients with reduced fibrinolytic capacity normalized their d-dimer scores after desmopressin stimulation. In accordance with our findings of unaltered d-dimer levels, analysis of all subjects by Andersen et al10 revealed no effect. It is possible that subgroups of patients with reduced fibrinolytic function might benefit from gemfibrozil treatment, but this needs to be confirmed in studies specifically designed to test this hypothesis.
Elevated plasma levels of lipoprotein(a) are associated with ischemic heart disease.31 This lipoprotein contains apo(a), which is structurally related to plasminogen.32 It has therefore been suggested that elevated plasma apo(a) levels might impair fibrinolysis. The present study showed no alteration of plasma apo(a) levels during gemfibrozil treatment, nor were any associations found between fibrinolytic variables and apo(a) levels.
It may be argued that concomitant medication and the inclusion of patients both with and without cardiovascular disease could have limited the interpretation of the study results. However, all concomitant medication was kept constant throughout the trial, and the patients served as their own controls. Furthermore, the studied population is representative of the patient category that receives gemfibrozil therapy. Another possible limitation of the study is its crossover design, with inherent risks of carryover from the first to second treatment period. This possibility, however, seems very unlikely to have influenced our results as our treatment periods were long (2.5 to 3 months). Furthermore, the statistical evaluation showed no signs of carryover effects.
In conclusion, gemfibrozil given to men with combined hyperlipoproteinemia influences plasma lipoprotein levels favorably without influencing the fibrinolytic system or fibrinogen levels. In addition, the dynamic regulation of fibrinolysis after mental stress or venous occlusion did not improve during gemfibrozil treatment. Insulin levels were not found to be a confounder with regard to the lack of gemfibrozil effect on fibrinolysis. Our results indicate that the relationship between triglycerides and PAI-1 is complex and that mechanisms other than improved fibrinolysis may explain the primary preventive effects of gemfibrozil on coronary heart disease.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|PAI-1||=||plasminogen activator inhibitor–1|
|TPA||=||tissue-type plasminogen activator|
This study was supported by grants from Parke-Davis, the Swedish Heart-Lung Foundation, the Swedish Medical Research Council (5930, 7137, and 5193), the Thuring Foundation, the Swedish Society of Medicine, King Gustaf V and Queen Victoria’s Foundation, and the Karolinska Institute. We are indebted to Maud Daleskog, Maj-Christina Johansson, Lilian Larsson, Tova Dahlström, and Marie Haegerstrand-Björkman for expert technical assistance and to Pia Gustafsson for excellent patient care.
Wiman B, Hamsten A. Decreased fibrinolytic activity in thrombotic disease: an update. In: Samama MM. Seghatchian MJ, eds. Hypercoagulable States: Biological Aspects and Clinical Management. Boca Raton, Fla: CRC Press: In press.
Mussoni L, Mannucci L, Sirtori M, Camera M, Maderna P, Sironi L, Tremoli E. Hypertriglyceridemia and regulation of fibrinolytic activity. Arterioscler Thromb. 1992;12:19-27.
Frick MH, Elo O, Happa 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. Helsinki Heart Study: primary-prevention trial with gemfibrozil in middle-age men with dyslipidemia. N Engl J Med. 1987;317:1237-1245.
Keber I, Lavre J, Suc S, Keber D. The decrease of plasminogen activator inhibitor after normalization of triglycerides during treatment with fibrates. Fibrinolysis. 1994;8:57-59.
Frankenhaeuser M, Mellis I, Rissler A, Björkvall C, Patkai P. Catecholamine excretion as related to cognitive and emotional reaction patterns. Psychosom Med. 1968;30:109-120.
Carlsson K. Lipoprotein fractionation. J Clin Pathol. 1973;26(suppl 5):32-37.
Lopes-Virella MF, Stone P, Ellis S, Colwell JA. Cholesterol determination in high density lipoproteins separated by three different methods. Clin Chem. 1977;23:882-884.
Hjemdahl P. Catecholamine measurements in plasma by high-performance liquid chromatography with electrical detection. Methods Enzymol. 1987;142:521-534.
Jones B, Kenward MG. Design and Analysis of Cross-over Trials. New York, NY: Chapman and Hall; 1989.
Paramo JA, Colucci M, Collen D, van de Werf F. Plasminogen activator inhibitor in the blood of patients with coronary artery disease. BMJ.. 1985;291:573-574.
Cortellaro M, Cofrancesco E, Boschetti C, Mussoni L, Donati MB, Cardillo M, Catalano M, Gabrielli L, Lombardi B, Specchia G, Tavazzi L, Tremoli E, Pozzoli E, Turri M, for the PLAT Group. Increased fibrin turnover and high PAI-1 activity as predictors of ischemic events in atherosclerotic patients. Arterioscler Thromb. 1993;13:1412-1417.
Juhan-Vague I, Vague P, Alessi MC, Badier C, Valadier J, Aillaud M, Atlan C. Relationships between plasma insulin, triglyceride, body mass index, and plasminogen activator inhibitor 1. Diabetes Metab. 1987;13:331-336.
Sundell IB, Dahlgren S, Rånby M, Stenling R, Nilsson TK. Reduction of elevated plasminogen activator inhibitor levels during modest weight loss. Fibrinolysis. 1989;3:51-53.
Juhan-Vague I, Thompson SG, Jespersen J, for the ECAT Angina Pectoris Study Group. Involvement of the hemostatic system in the insulin resistance syndrome, a study of 15 000 patients with angina pectoris. Arterioscler Thromb. 1993;13:1865-1877.
Elkeles RS, Chakrabarti R, Vickers M, Stirling Y, Meade T. Effect of treatment of hyperlipidaemia on haemostatic variables. BMJ. 1980;281:973-974.