Fibrinogen Is a Predictor of Mortality in Coronary Heart Disease Patients
Abstract Results of epidemiological studies have indicated that fibrinogen is an important primary cardiovascular risk factor. The role of fibrinogen as a predictor of mortality in coronary heart disease (CHD) patients is unclear. We investigated the association between fibrinogen and mortality in a large cohort of CHD patients screened for participation in a secondary prevention clinical trial. Of the total investigated, 3092 men who were not included in the trial and for whom vital status was known were followed up for a mean period of 3.2 years. In 54.4% of the 111 men who died, mortality was attributed to CHD. Mean baseline plasma fibrinogen levels were 29.4 mg/dL higher in patients who died than in the survivors. All-cause and CHD mortality rates increased with increasing fibrinogen levels. This relationship was also demonstrated within categories of the primary variables predicting mortality in these patients. The contribution of fibrinogen to CHD and all-cause mortality was assessed by multivariate analysis adjusting for age, CHD severity, and comorbidity. Risks of CHD and all-cause mortality for patients in the highest fibrinogen tertile were 1.67 and 1.75, respectively, relative to patients in the lowest tertile, and an increase of about 1 SD of plasma fibrinogen level (75 mg/dL) was found to increase risk of CHD and all-cause mortality by 29% and 31%, respectively. These results indicate clearly that fibrinogen level is associated with significantly increased mortality in CHD patients. Implementation of a standardized measuring method is required to allow assessment of risk in CHD patients on the basis of fibrinogen levels.
See “Appendix” for list of participating centers, principal investigators, and physicians.
- Received July 12, 1995.
- Accepted December 19, 1995.
The role of fibrinogen as a primary cardiovascular risk factor is well established and has been demonstrated by a number of prospective epidemiological studies of healthy individuals1 2 3 4 5 6 . In a meta-analysis of six prospective studies,7 the odds of sustaining a cardiovascular event in healthy persons with a fibrinogen level in the highest tertile were 2.3 times as high as in those with fibrinogen levels in the lowest tertile. An association between fibrinogen and mortality attributed to cardiovascular disease was reported as early as 1980 by the Northwick Park Heart study of 1510 men, most of whom were free of ischemic heart disease at recruitment.1 These findings were later reinforced by a report of 10 years of follow-up.8
One question that remains to be answered pertains to the importance of fibrinogen as a predictor of mortality in patients who already manifest cardiovascular disease, CHD in particular. A recent study9 reported that an increase of 100 mg/dL of fibrinogen in patients with stable intermittent claudication predicted a nearly twofold increase in the probability of death within the next 6 years. Another study of 1716 men 6 months after an index MI10 reported a trend of increasing odds of ischemic events with increasing fibrinogen levels during 2.5 years of follow-up.
The screening of patients for recruitment to a clinical secondary prevention trial enabled us to investigate the association between fibrinogen and mortality in a large cohort of CHD patients.
Approximately 15 500 CHD patients with previous MI (0.5 to 5 years before inclusion in the clinical trial) and/or stable angina pectoris (within 2 years preceding inclusion), most of them aged 45 to 74, were screened for participation in the BIP study between February 1990 and January 1993. The major exclusion criteria were: permanent pacemaker implantation, cerebrovascular disease, chronic renal or liver disease, severe chronic obstructive pulmonary disease, severe peripheral vascular disease, severe noncoronary heart disease, insulin-dependent diabetes, malignant disease, and estrogen therapy for women. The complete list has been published.11
Patients with total cholesterol ≤7.0 mmol/L (270 mg/dL), HDL cholesterol ≤1.17 mmol/L (45 mg/dL), and serum triglyceride ≤3.39 mmol/L (300 mg/dL) at the time of the first screening visit were invited to a second visit after a 2-month diet, at which time the concentration of fibrinogen in the plasma was determined.
Mortality data were obtained through September 1994 from the Israeli Population Registry for 14 655 screened patients for whom the identification details were matched successfully to the registry records. Cause of death was coded according to ICD/9 codes. Fibrinogen measurements were available for 6561 of these patients. Since the sample of women was small (745 women, of whom 28 died) the analysis was restricted to men. To avoid the possible modifying effect of bezafibrate treatment on the association between fibrinogen and mortality, we report on 3092 patients who were not included in the BIP study and for whom vital status was ascertained.
Cooled blood samples, collected in the 18 participating centers using standard equipment and procedures, were transferred to the central laboratory of the study. All analysis was performed on a Boehringer Hitachi 704 random access analyzer using Boehringer diagnostic kits. Fibrinogen was determined in citrated plasma using a kinetic method, as described by Hemker et al.12 Evaluation of triglycerides was done by determining total triglycerides (GPO-PAP high-performance method) and subtracting free glycerol (determined by using an enzymatic kit manufactured by Sigma). HDL cholesterol was determined by precipitation of LDL and VLDL with phosphotungstate reagent and determination of cholesterol. Cholesterol was determined by the CHOD-PAP method (enzymatic colorimetric test). LDL cholesterol was approximated by the formula of Friedewald et al13 for patients with serum triglyceride ≤4.5 mmol/L (400 mg/dL).
Drifts in fibrinogen levels were detected twice during data collection, in the 2nd and 17th months of follow-up, owing to change of analytical kits. Therefore, adjustment of fibrinogen level was performed to ensure comparability of measurement across the screening period. Since the kit change resulted in a linear shift of fibrinogen values, adjustment was done by reanalyzing frozen plasma collected before the kit change with the kits used routinely at the time of the reanalysis and applying linear regression analysis to obtain an estimation of old values on the new scale. The adjustment formula for the first kit period is A=80.42+0.56×O and for the second kit period, A=32.25+0.81×O, where A represents the adjusted level and O the original measurement of fibrinogen. Some fluctuation in monthly means of measured fibrinogen levels were observed across the data collection period. Means ranged between 296 and 370 mg/dL (SD=14 mg/dL).
Determination of Additional Parameters
Diabetes and hypertension (defined by history), as well as smoking habits, were recorded on the basis of self-reporting by the patient during an interview held with a study physician. Clinical findings were assessed by the interviewing physician. Criteria for the diagnosis of MI and angina pectoris have been published.11
Fibrinogen was nearly symmetrically distributed (mean=346±77, median=336), with tertile cut points of 308.3 and 367.7. Age-adjusted mortality rates per 1000 person-years were computed using a special SAS macro.16 Linear trend in unadjusted mortality rates in fibrinogen quintiles was tested using the Mantel-Haenszel χ2 statistic.
Multivariate analysis of mortality was performed using the Cox proportional hazard model (PHREG procedure) to account for differing lengths of follow-up and correlation with covariates. Fibrinogen was introduced into the model either as a continuous variable or by tertiles as two dummy variables, the bottom tertile being the reference group. The significance levels for entering and removing an explanatory variable were set at 0.15 and 0.10, respectively. The entry of fibrinogen tertiles into the model was forced when fibrinogen levels were introduced into the model as dummy variables. Forcing fibrinogen tertiles into the model was necessary for the inclusion of both dummy variables in the final model under the significance level for entering and removing variables into and from the model.
The mean age of the 3092 male CHD patients analyzed was 59 years (SD=7) at baseline. The majority of the patients (76%) had sustained an MI in the past, 58% reported suffering from angina pectoris, 26% were in functional capacity class II or higher, according to the New York Heart Association classification, and 3% had a history of peripheral vascular disease. In addition to the basic ischemic disease, 20% of patients were diabetics and 30% were hypertensive. Most of the patients (60%) had smoked in the past and stopped at least 3 months before baseline, and only 11% were active smokers at the beginning of follow-up. Among them, 79% had been smoking for 20 years or more. Mean lipid levels were: 5.77±0.88 mmol/L (223±34 mg/dL) for total cholesterol, 0.92±0.19 mmol/L (35.6±7.2 mg/dL) for HDL cholesterol, 4.02±0.83 mmol/L (155±32 mg/dL) for LDL cholesterol, and 1.82±0.93 mmol/L (161±82 mg/dL) for triglyceride.
The patients were followed up for a mean time of 3.2 years (follow-up time until death or September 1994 [analysis cutoff date] ranged between 0.15 and 4.6 years). Follow-up includes a total of 9884 person-years. Two hundred and four patients (6.6%) died during follow-up. For 111 men (54.4% of the deceased), the underlying cause of death was CHD (ICD/9 codes 410 to 414). Additional causes of death were: other heart disease (ICD/9 codes 415 to 429), 7.4%; cerebrovascular accident (ICD/9 codes 430 to 438), 2.0%; cancer (ICD/9 codes 140 to 209), 16.7%; and other causes, 11.3%. For 7.8%, the cause of death was unknown. Although differences in mean fibrinogen levels between patients grouped by the cause of death were not statistically significant, patients for whom the underlying cause of death was cancer exhibited the highest mean plasma fibrinogen levels in the screening examination (Table 1⇓), and 11 patients who died of lung malignancies had the highest baseline mean fibrinogen level when grouped separately.
Fig 1⇓ shows all-cause and CHD mortality in quintiles of fibrinogen levels. All-cause mortality increased from 15.1 in the bottom quintile to 33.4 in the highest quintile (test for linear trend: P<.0001). The rate of mortality attributed to CHD ranged between 8.1 in the lowest quintile and 17.4 in the highest one (test for linear trend: P=.0001). To demonstrate the persistent nature of the association between fibrinogen and mortality, all-cause mortality was examined by fibrinogen tertiles in categories of the primary variables predicting mortality (Fig 2⇓). Excess mortality in patients with high fibrinogen was observed in those with and without a history of MI, diabetes, and hypertension, as well as for smokers and nonsmokers, and by functional capacity classes according to the New York Heart Association classification. Higher mortality rates were also found in patients with fibrinogen in the highest tertile compared with the lowest tertile, within tertiles of total, HDL, and LDL cholesterol and triglyceride (Fig 3⇓).
Since smoking is one of the important modifiable variables determining fibrinogen level, we repeated these analyses, adjusting for smoking status and age. The resulting mortality rates were practically identical to those obtained after adjustment for age only.
The final step was to estimate the contribution of fibrinogen to CHD and all-cause mortality, adjusting for covariates and taking into consideration time elapsed between fibrinogen assessment and death. Stepwise multivariate analysis was performed using the proportional hazard method. The increased mortality risk associated with increased plasma fibrinogen was estimated, adjusting for age, total and HDL cholesterol, triglyceride (log transformation), history of MI, diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, angina pectoris, and smoking habits. Data on the above variables were available for 3047 of the 3092 patients.
Age, fibrinogen, diabetes, history of MI, and peripheral vascular disease were selected in a stepwise procedure for the final model for all-cause mortality. The relative risk for the latter associated with an increase of 75 mg/dL (about 1 SD) of plasma fibrinogen was 1.31 (95% CI: 1.16-1.47). Diabetes, fibrinogen, age, history of MI, and angina pectoris were included in the final model for CHD mortality, for which the relative risk associated with an increase of 75 mg/dL of fibrinogen level was 1.29 (95% CI: 1.10-1.51). The same variables were considered when tertiles of fibrinogen were included as two dummy variables. Risks for mortality relative to the lowest tertile are presented in Table 2⇓. Similar results were obtained when smoking was forced into the model. The relative risks of all-cause mortality associated with the middle and highest fibrinogen tertiles were 1.45 (95% CI: 1.0-2.11) and 1.66 (95% CI: 1.15-2.39), respectively. Forcing HDL and total cholesterol into the model as well yielded similar estimates.
Studies in healthy individuals have shown that plasma fibrinogen is a strong predictor for cardiovascular disease and mortality. CHD patients were reported to have higher fibrinogen levels than healthy individuals.17 Since reports in CHD patients are scarce, it is not clear whether fibrinogen has a similar prognostic importance in the latter population or whether high fibrinogen is merely a marker of the existing disease. The results presented here show that high plasma fibrinogen characterizes CHD patients who are more likely to die prematurely from ischemic heart disease. Such an association with all-cause mortality was not previously reported in CHD patients. In addition, we have observed a parallel increase in coronary mortality in these patients.
Martin and coworkers10 studied the influence of platelet size and other hematological variables (including fibrinogen) on the outcome in male MI survivors under 70 years of age (diabetic patients and those who needed cardiac surgery were excluded from the study). The authors noted a consistent trend of increasing odds of ischemic events and death with increasing fibrinogen. Our study included MI survivors as well as patients with angina pectoris. Twenty percent of patients had a history of diabetes. Results show that mortality rates increased linearly with increasing fibrinogen quintiles. This relationship is in agreement with findings in prospective studies of healthy individuals.4
Previous findings in patients screened for the BIP study suggested that fibrinogen is associated to a small extent with other risk factors for mortality in these patients.18 Despite these associations, which are demonstrated for healthy individuals as well,19 the correlation between plasma fibrinogen level and mortality did not merely reflect the relationship with other risk factors or mark the extent of the underlying disease. The bivariate analyses showed that the association between fibrinogen tertiles and mortality was preserved within categories of the major risk factors for mortality and thus support the independent predictive role of fibrinogen. Adjusting for these risk factors in a multivariate analysis reinforces this conclusion. An exception was the absence of increasing mortality with increasing fibrinogen levels for men in the highest tertile of triglycerides. This could be typical of multiple-comparison situations, and we do not have a readily available explanation for this posthoc finding.
The highest fibrinogen levels were found in patients who died of cancer; in particular, lung malignancies. In a study of 297 men initially free of overt CHD,2 19 men in whom a malignant disease was diagnosed during 7.5 years of follow-up had had higher fibrinogen levels than individuals who remained disease free. Smoking, which was more frequent among cancer patients, was suggested as a reason. Higher plasma fibrinogen level in cancer patients may also be caused by a nonspecific inflammatory reaction to tumor cells.
Several plausible mechanisms were suggested to explain how raised fibrinogen levels promote CHD, eg, involvement in occlusive and mural thrombus formation, platelet aggregation, atherogenesis, and determination of blood viscosity. These mechanisms may be relevant to aggravation of an already-existent disease. In addition, elevated levels of d-dimer, a primary degradation product of cross-linked fibrin, were reported to be associated with subsequent CHD development among peripheral vascular disease patients20 and among apparently healthy individuals.21 Our findings and this indication of fibrinolysis suggest an activation of the coagulation cascade and fibrinolysis in an ongoing process of formation and degradation of fibrin, contributing to progression of both atherosclerosis and thrombosis.
The relationship between fibrinogen and CHD in women has been dealt with only in the general population by the Framingham study.5 Because of the small sample, women were not included in this analysis. Albeit higher fibrinogen levels were found in women than in men in our study sample,18 women had a lower mortality rate (3.8% of women died compared with 6.6% of men). Longer follow-up will allow us to draw conclusions regarding women as well.
These results highlight the advisability of adding the assessment of plasma fibrinogen concentration to the biochemical profile derived for CHD patients. They emphasize the importance of finalizing and implementing an internationally standardized method for measuring plasma fibrinogen so as to allow comparability of international data. This would further permit the determination of fibrinogen concentration consistent with low, intermediate, and high risk, as established for blood cholesterol. Research in large, unselected groups of patients with CHD is required in order to establish the quantitative nature of the association between plasma fibrinogen, recurrent infarction, CHD, and stroke mortality, as well as mortality from other causes, in men and women with clinically manifest CHD.
Whether lowering fibrinogen levels in CHD patients will result in reduction of mortality in these patients has yet to be discovered. The ongoing BIP study, designed to test whether elevation of HDL cholesterol and simultaneous reduction in triglyceride and fibrinogen levels via administration of bezafibrate will reduce mortality and coronary events, is likely to shed light on this issue.
Selected Abbreviations and Acronyms
|BIP||=||Bezafibrate Infarction Prevention study|
|CHD||=||coronary heart disease|
|ICD/9||=||International Classification of Disease, 9th version|
Participating Centers and Committee Membership
Jacob Agmon, MD; Solomon Behar, MD; Daniel Brunner, MD (Chairman); Avraham Caspi, MD; Uri Goldbourt, PhD; Eran Graff, PhD; Elieser Kaplinsky, MD; Yehezkiel Kishon, MD; Henrietta Reicher-Reiss, MD; and Joshua Waysbort, MD.
Participating Centers, Principal Investigators, and Physicians
Assaf Harofe Hospital, Zrifin: Zwi Schlesinger, MD; and Aharon Fridensohn, MD.
Barzilai Medical Center, Ashkelon: Leonardo Reisin, MD; and Jamal Jafari, MD.
Beilinson Medical Center, Petach Tikva: Samuel Sclarovsky, MD; Yaakov Friedman, MD; and Bruno Ostfeld, MD.
Bnei-Zion Medical Center, Haifa: Edward G Abinader, MD; and Shmuel Rauchfleish, MD.
Carmel Hospital, Haifa: Abraham Palant, MD; and Hanan Schneider, MD.
Central HaEmek Hospital, Afula: Tiberio Rosenfeld, MD; and Suleiman Khalid, MD.
Edith Wolfson Medical Center, Holon: Yehezkiel Kishon, MD; and Rene Rotzak, MD.
Hasharon Hospital, Petach-Tikva: Izhar Zahavi, MD; and Janash Vitrai, MD.
Hillel Yaffe Hospital, Hadera: Benyamin Pelled, MD; and Joseph Pardu, MD.
Ichilov Hospital, Sorasky Medical Center, Tel-Aviv: Shlomo Laniado, MD; Libi Sherf, MD; Shimon Braun, MD; and Yemima Eschar, MD.
Kaplan Hospital, Rehovot: Avraham Caspi, MD; Alexander Arditi, MD; and Shulamit Botwin, MD.
Meir Hospital, Sapir Medical Center, Kfar Saba: Daniel David, MD; and Joel Arbel, MD.
Naharia Hospital, Naharia: Nathan Roguin, MD; and Alicia Glusman, MD.
Rambam Medical Center, Haifa: Walter Markiewicz, MD; and Diav Motlak, MD.
Rivka Ziv Hospital, Tzfad: Alon Marmour, MD; and Michael Flich, MD.
Shaare Zedek Medical Center, Jerusalem: Monty Zion, MD; and Jonathan Balkin, MD.
Sheba Medical Center Heart Institute, Tel Hashomer: Babeth Rabinowitz, MD; and Eddy Barasch, MD.
Soroka Medical Center, Be’er Sheba: Natalio Kristal, MD; and Noah Liel, MD.
Jacob Agmon, MD; Yisrael Bar-Yehuda; Solomon Behar, MD (Medical Director); Daniel Brunner, MD; Uri Goldbourt, PhD; Elieser Kaplinsky, MD; Lori Mandelzweig, MPH; and Henrietta Reicher-Reiss, MD.
Daniel Brunner, MD; Eran Graff, PhD (director); Sara Schwarz, Msc.
Stone MC, Thorp J. Plasma fibrinogen: a major coronary risk factor. J R Coll Gen Pract. 1985;35:564-569.
Yarnell JWG, Baker IA, Sweetnam PM, Bainton D, O’Brien JR, Whitehead PJ, Elwood PC. Fibrinogen, viscosity, and white blood cell count are major risk factors for ischemic heart disease. Circulation. 1991;83:836-844.
Heinrich J, Balleisen L, Schulte H, Assmann G, van de Loo J. Fibrinogen and factor VII in the prediction of coronary risk: results from the PROCAM study in healthy men. Arterioscler Thromb. 1994;14:54-59.
Goldbourt U, Behar S, Reicher-Reiss H, et al for the Bezafibrate Infarction Prevention Study Group. Rationale and design of a secondary prevention trial of increasing serum HDL cholesterol and reducing triglycerides after myocardial infarction in patients with clinically manifest atherosclerotic heart disease (the Bezafibrate Infarction Prevention study). Am J Cardiol. 1993;71:909-915.
Friedewald WT, Levy RI, Frederickson DS. Estimation of the concentration of low density lipoprotein cholesterol in plasma, without use of preparative ultracentrifuge. Clin Chem. 1972;18:499-502.
SAS/STAT User’s Guide, Version 6. 4th ed. Cary, NC: SAS Institute Inc; 1987;1:943 and 1987;2:620.
SAS/STAT software: changes and enhancements, release 6.07. Cary, NC: SAS Institute Inc; 1992:620. SAS technical report P-229.
Lee AJ, Lowe GDO, Woodward M, Tunstall-Pedoe H. Fibrinogen in relation to personal history of prevalent hypertension, diabetes, stroke, intermittent claudication, coronary heart disease, and family history: the Scottish Heart Health Study. Br Heart J. 1993;69:338-342.
Barasch E, Benderly M, Graff E, Behar S, Reicher-Reiss H, Caspi A, Peled B, Reisin L, Roguin N, Goldbourt U. Plasma fibrinogen levels and their correlates in 6457 coronary heart disease patients: the Bezafibrate Infarction Prevention (BIP) study. J Clin Epidemiol. 1996;48:757-765.
Lee AJ, Smith WCS, Lowe GDO, Tunstall-Pedoe H. Plasma fibrinogen and coronary risk factors: the Scottish Heart Study. J Clin Epidemiol. 1990;16:83-88.
Ridker PM, Hennekens CH, Cerskus A, Stampfer MJ. Plasma concentration of cross-linked fibrin degradation product (d-dimer) and the risk of future myocardial infarction among apparently healthy men. Circulation. 1994;90:2236-2240.