OUP user menu

Plasma fibrinogen associates independently with total and cardiovascular mortality among subjects with normal and reduced kidney function in the general population

A.G. Stack, U. Donigiewicz, A.A. Abdalla, A. Weiland, L.F. Casserly, C.J. Cronin, H.T. Nguyen, A. Hannigan
DOI: http://dx.doi.org/10.1093/qjmed/hcu057 hcu057 First published online: 14 March 2014

Abstract

Background: The contribution of novel risk factors to mortality in chronic kidney disease remains controversial.

Aim: To explore the association of plasma fibrinogen with mortality among individuals with normal and reduced kidney function.

Methods: We identified 9184 subjects, age 40 and over from the Third National Health and Nutrition Examination Survey (1988–94) with vital status assessed through 2006. Plasma fibrinogen was modeled as continuous variable and in quartile groups (0 to <7.7, 7.7 to <9.0, 9.0 to <10.5 and ≥10.5 µmol/l) with total and cardiovascular mortality across categories of glomerular filtration rate (eGFR); <60, 60–90, >90 ml/min/1.73 m2 using Cox regression.

Results: In multivariate analysis, the adjusted hazard ratio (HR) per 1 µmol/l (34 mg/dl) increase in fibrinogen was 1.07 [95% confidence interval (CI) 1.04–1.09] for total mortality and 1.06 (95% CI 1.03–1.09) for cardiovascular mortality. The adjusted HR for total mortality was 1.05 (1.01–1.09) for subjects with eGFR 60–90 ml/min/1.73 m2 and 1.06 (1.02–1.10) for subjects with eGFR <60 ml/min/1.73 m2. Subjects in the highest quartiles within each eGFR category; >90, 60–90 and <60 ml/min/1.73 m2 experienced HRs of 1.45 (95% CI 1.03–2.03), 1.35 (95% CI 1.00–1.83) and 1.72 (95% CI 1.14–2.58), respectively, compared with subjects in the lowest quartile group. The patterns were similar for cardiovascular mortality.

Conclusions: Plasma fibrinogen associates with mortality among subjects with mild to moderate kidney impairment as it does in subjects with normal kidney function and should be considered a therapeutic target for cardiovascular risk reduction.

Introduction

Chronic kidney disease (CKD) is a major predictor of total and cardiovascular mortality in the general population.1 The unadjusted relative risks for mortality are between 1.4 and 3.7 higher for CKD patients than those without CKD and the absolute risk of death increases exponentially with decreasing renal function.1 The causes for this excess cardiovascular risk are not fully understood although it is believed that both traditional Framingham risk factors and novel risk factors have contributory roles.2–5 Although it is clear that many traditional risk factors are present in higher concentrations in CKD, cardiovascular risk equations developed in the general population consistently underestimate cardiac event rates in subjects with kidney impairment suggesting that conventional risk factors do not adequately explain the excess risk of cardiovascular events.6 Consequently, much attention has focused on the potential contributions of novel cardiovascular risk factors such as inflammatory and prothrombotic factors that accumulate with worsening kidney function.7–10

Plasma fibrinogen is an important component of the coagulation cascade and a major determinant of blood viscosity and blood flow.11–14 The evidence from epidemiological studies in the general population suggests that elevated plasma fibrinogen concentrations are associated with an increased risk of cardiovascular events, including ischaemic heart disease, stroke.15 In contrast, the evidence in CKD cohorts is less convincing. Although few studies have demonstrated positive associations between plasma fibrinogen and cardiovascular mortality among patients undergoing dialysis treatment,16,17 there is conflicting evidence among patients with lesser degrees of kidney impairment especially within the general population. A prospective study of 128 patients with stage 3–stage 4 CKD and followed for 5.5 years found that elevated levels predicted future cardiovascular events.18 In contrast, a much larger cohort of 5808 older community adults followed for 8.6 years failed to find a significant association between fibrinogen and cardiovascular death among the subset with CKD.19 Although these studies have contributed to our knowledge base, there is residual doubt as to whether plasma fibrinogen associates with future cardiovascular risk among subjects with mild to moderate kidney impairment.

The purpose of this study was (i) to explore the nature and magnitude of the relationships between plasma fibrinogen and mortality in subjects with normal and impaired kidney function in the general population and (ii) to determine whether relationships between serum fibrinogen and cardiovascular mortality could be explained by Framingham risk factors and inflammatory markers.

Methods

NHANES III study

The Third National Health and Nutrition Examination Survey (NHANES III) was a population-based survey conducted by the National Center for Health Statistics that assessed the health status of a representative sample of non-institutionalized persons living in the USA between 1988 and 1994.20 A complex stratified multistage probability design was used with over-sampling of elderly and minority populations. A standardized questionnaire was administered and detailed physical examination conducted at a mobile examination centre or at the participant’s home. The survey was approved by the Institutional Review Board of the National Center for Health Statistics, and informed consent was obtained before participation. The vital status of almost all NHANES III participants (99.9%), 17 years of age or older, was determined through matching to the National Death Index through 31 December 2006.21

Sample

We identified adults age 20 or older who had serum creatinine measurements performed in NHANES III (n = 15 575). From this sample, we restricted the analysis to subjects with data on plasma fibrinogen levels (age 40 and over) and who had vital status assessed to 31 December 2006 (n = 9184).

Clinical data

The NHANES III provided data that were collected from face-to-face interviews and direct physical examinations. Information was obtained on sociodemographic characteristics, medical and family history. The presence of comorbid conditions was based on self report that required a physician’s diagnosis. Trained observers measured blood pressure three times during the home interview and three times during the subsequent evaluation at the mobile examination centre using a standard protocol. Blood pressure for individual participants was calculated as the average of all available systolic and diastolic readings. Physical examination included data on height, weight and the average of at least three blood pressure readings. Body mass index was calculated as weight/height2 (kg/m2).

Baseline measurements

Blood samples were obtained from non-fasting persons and frozen serum sent to the Centers for Disease Control and Prevention for analysis.22 Plasma fibrinogen was measured using enzyme assay methods with Coag-A-Mate XC plus (Organon-Teknika, Alamogordo, NM). Serum C-reactive protein levels were measured using latex-enhanced nephelometry (Behring Nephelometer Analyzer System; Behring Diagnostics, Somerville, NJ). Serum ferritin was measured with the BioRad Quantimmune IRMA kit (BioRad Laboratories, Hercules, CA). Serum total cholesterol was measured enzymatically using a commercially available reagent mixture (Cholesterol/HP; Boehringer Mannheim Diagnostics). Serum creatinine concentrations were measured by the modified kinetic Jaffe reaction using a Hitachi 737 analyzer (Boehringer Mannheim Corp., Indianapolis, IN) and recalibrated to the Cleveland Clinic to ensure validity of the results.23 Glomerular filtration rate was estimated from the abbreviated Modification of Diet in Renal Disease (MDRD) Study and the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations.24,25 Participants were divided into groupings of estimated GFR (≥90, 60–89, and <60 ml/min per 1.73 m2).

Assessment of all-cause and cardiovascular mortality

The occurrence of death and cause of death for all cohort participants were provided through linkage with the National Death Index.21 and vital status was available up to 31 December 2006. Cardiovascular causes of death were identified from the International Classification of Diseases, 10th Revision (ICD-10) diagnosis codes in the NHANES-linked mortality files and included: deaths from acute myocardial infarction (121–122), other acute ischemic heart disease (124), atherosclerotic cardiovascular disease (125.0), all other forms of chronic ischemic heart disease (120, 125.1, 125.9) and cerebrovascular disease (160–169). Deaths were analyzed for all causes and cardiovascular causes (ICD-10 codes 120–169).

Statistical analysis

Plasma fibrinogen concentration was modelled as a continuous variable and in quartile groups. Subjects were stratified into quartiles of baseline fibrinogen and their characteristics were compared across groups. Similarly, baseline characteristics of the population were also compared across glomerular filtration rate (eGFR) categories (<60, 60–89 and ≥90 ml/min/1.73 m2). For continuous variables, differences across quartiles were tested with analysis of variance whereas for dichotomous variables comparisons were conducted using the chi-square test and all comparisons were weighted according to the complex sampling strategy.

Total and cardiovascular mortality rates were calculated for the entire cohort expressed as deaths per 1000 person-years. Cox proportional hazard regression was used to model relationships of plasma fibrinogen as a continuous variable and in quartiles groups with total and cardiovascular mortality adjusting for baseline characteristics. The assumption of proportionality of the hazards ratios was evaluated by plots of Schoenfeld residuals. Relationships between plasma fibrinogen and mortality were explored in sequentially adjusted models that included age (continuous), sex, race and Framingham risk factors; history of diabetes and hypertension, measured blood pressure at examination, family history of coronary disease, physical inactivity, cholesterol/HDL ratio and smoking behaviour (never, past, current). To determine whether the association of plasma fibrinogen with mortality differed by level of kidney function, we tested for an interaction between eGFR and plasma fibrinogen with mortality. A significant interaction was defined as P < 0.01. We then stratified the sample by eGFR category (<60, 60–89 and ≥90 ml/min/1.73 m2) and repeated the analysis with similar adjustment for confounding. The functional relationship between plasma fibrinogen and death was further examined using restricted cubic spline functions with five knots placed at the 5th, 25th, 50th, 75th and 95th percentiles of fibrinogen. The assumption of a linear relationship between plasma fibrinogen and the log hazard of mortality was tested using a Wald chi-square test and adjusted for age, sex and Framingham factors using Cox regression. For each covariate, the unadjusted hazard ratio (HR) and adjusted HR of death were calculated with corresponding 95% confidence intervals (CIs). Weighted Cox regression was used to account for the complex survey design, the unequal probability of subject selection and non-response rates. Model fit was assessed using the Taylor method and −2 log likelihood ratio (SAS v 9.3, New York, SAS Institute, Cary, NC, USA).

Results

Baseline characteristics of the population

The mean characteristics of all study participants age 40 and over are illustrated in Table 1. The weighted mean age of the study population and standard error (±SE) was 57.4 years (±0.41), 81.2% were White, 8.9% were Black and 3.5% were Mexican-American. The mean plasma fibrinogen was 9.0 (±0.08) µmol/l (to convert to mg/dl multiply by 34.0).

View this table:
Table 1

Characteristics of study participants by quartiles of fibrinogena

Fibrinogen (quartiles)
Patient characteristicsRespondentsEntire cohortFirstSecondThirdFourth
(0 < 7.7)(7.7 to <9.0)(9.0 to <10.5)(≥10.5)
Demographics
    Age at interview (years)b918457.4 (0.41)53.3 (0.42)56.1 (0.59)58.3 (0.67)61.7 (0.52)
    Sex9184
        % Men438746.5 (0.66)53.9 (1.69)47.7 (1.85)44.4 (1.78)40.2 (1.33)
        % Women479753.5 (0.66)46.1 (1.69))52.3 (1.85)55.6 (1.78)59.8 (1.33)
    Race/ethnicity
        %White-non Hispanic471081.2 (1.20)83.5 (1.32)81.4 (1.71)80.5 (1.60)79.5 (1.74)
        % Black non-Hispanic20848.9 (0.54)7.3 (0.55)7.9 (0.81)8.8 (0.71)11.5 (0.89)
        % Mexican American20423.5 (0.26)3.3 (0.44)3.5 (0.32)3.7 (0.35)3.4 (0.36)
        % Other3486.5 (0.91)6.0 (0.95)7.2 (1.26)6.9 (1.20)5.6 (1.20)
Clinical condition (% present)
    History of ischaemic heart disease91799.2 (0.44)5.9 (0.78)7.7 (0.88)9.9 (0.74)13.2 (0.85)
    History of myocardial infarction90856.2 (0.39)3.7 (0.56)4.3 (0.58)6.9 (0.66)9.6 (0.87)
    History of angina66916.1 (0.45)4.1 (0.67)6.6 (0.89)6.1 (0.88)7.4 (0.83)
    Chronic bronchitis91837.9 (0.39)5.1 (0.70)6.2 (0.75)7.6 (0.78)12.8 (0.88)
    Congestive heart failure91633.7 (0.23)1.9 (0.26)2.2 (0.27)3.6 (0.57)7.0 (0.60)
    Stroke91793.3 (0.25)2.1 (0.46)2.1 (0.42)3.9 (0.52)5.3 (0.48)
    Diabetes (by history)91748.2 (0.37)5.2 (0.77)5.8 (0.71)8.6 (0.77)13.1 (0.66)
    Hypertension (by history)914134.2 (0.83)25.9 (1.46)29.6 (1.49)35.8 (1.46)45.4 (1.22)
 JNC VII staging of hypertension
        Normal < 120/80 %594068.8 (0.89)71.2 (1.75)67.4 (1.88)67.1 (1.74)69.6 (1.82)
        Pre-hypertension 120/80 %227025.2 (0.69)23.0 (1.60)27.1 (1.65)26.0 (1.37)24.7 (1.56)
        Stage I 140/90 %5385.2 (0.37)5.3 (0.82)5.0 (0.77)5.8 (0.67)4.5 (0.64)
        Stage II 160/100 %1140.80 (0.14)0.49 (0.18)0.46 (0.17)1.11 (0.29)1.16 (0.31)
    Family history of myocardial infarction < 50 yrs899715.3 (0.62)15.6 (1.77)13.4 (1.62)16.1 (1.36)16.2 (1.14)
    Tobacco use9184
        Current smokers %198522.9 (0.90)18.5 (1.59)19.7 (1.53)24.1 (1.73)29.2 (1.32)
        Former smokers %304835.0 (0.79)36.8 (1.55)35.9 (1.41)33.8 (1.76)33.7 (1.09)
        Never smokers, %415142.1 (0.84)44.8 (1.90)44.4 (1.73)42.1 (2.01)37.0 (1.39)
    Physically inactive (%)918417.9 (0.82)13.4 (1.10)14.9 (1.30)17.8 (1.19)25.4 (1.29)
    Body mass index (kg/m2) (mean ± SE)915227.3 (0.12)26.1 (0.14)26.8 (0.13)27.8 (0.20)28.5 (0.22)
Laboratory variables
    Plasma fibrinogen (µmol/l)91849.0 (0.08)6.4 (0.04)8.0 (0.01)9.3 (0.02)12.2 (0.10)
    Total cholesterol (mmol/l)91605.64 (0.02)5.4 (0.04)5.6 (0.04)5.7 (0.03)5.8 (0.03)
    HDL cholesterol (mmol/l)90871.3 (0.01)1.4 (0.02)1.3 (0.01)1.3 (0.02)1.3 (0.01)
    Total cholesterol/HDL ratio90844.7 (0.05)4.4 (0.07)4.6 (0.09)4.8 (0.07)5.0 (0.06)
    C-reactive protein (nmol/l)91750.47 (0.01)0.27 (0.01)0.33 (0.01)0.41 (0.01)0.87 (0.04)
    Ferritin (ng/ml)9160150.6 (2.81)150.2 (5.3)141.3 (4.9)150.0 (3.6)161.0 (5.28)
    Serum creatinine (µmol/l)918497.7 (0.39)95.7 (0.55)1.10 (0.48)97.2 (0.55)101.0 (0.89)
    eGFR MDRD (ml/min/1.73 m2)c918489.5 (0.51)93.9 (0.66)89.8 (0.71)88.8 (0.62)85.9 (0.85)
    eGFR CKD-EPI ((ml/min/1.73 m2)d918487.5 (0.48)92.1 (0.43)88.6 (0.63)86.8 (0.67)82.4 (0.76)
  • aThe proportions and means are weighted according to the probability sampling strategy and account for non-response and missingness. Conversion of µmol/l to mg/dl multiply by 34.0. bMean values are reported with standard errors (±SE) in parenthesis. cEstimated glomerular filtration rate (ml/min per 1.73 m2) was based on the MDRD Study equation.24 dEstimated glomerular filtration rate (ml/min per 1.73 m2) was based on the CKD-EPI equation.25

Characteristics of population by quartile of plasma fibrinogen

Higher quartiles of plasma fibrinogen were associated with an increasing prevalence of common cardiovascular conditions as shown in Table 1. Plasma fibrinogen concentrations increased from a mean of 6.4 µmol/l in the lowest quartile to 12.2 µmol/l in the highest quartile. Higher plasma concentrations correlated with higher C-reactive protein levels and serum ferritin levels (P < 0.01) and lower eGFR rates (P < 0.001).

Plasma fibrinogen levels and level of kidney function

Table 2 describes the characteristics of the study population by eGFR category. Subjects categorized as having eGFR <60 ml/min had the highest prevalence of cardiovascular conditions and Framingham risk factors, and highest concentrations of plasma fibrinogen.

View this table:
Table 2

Characteristics of study participants by level of kidney function

GFR categories (quartiles)
Patient characteristicsRespondentsEntire cohortFirstSecondThird
(n)% or Mean ± SEa(≥90)(60 to <90)(0 < 60)
Demographics
    Age at interview (years)b918457.4 (0.41)53.6 (0.38)58.9 (0.41)72.2 (0.64)
    Sex
        % Men438746.5 (0.66)48.2 (0.92)46.1 (1.09)38.4 (2.7)
        % Women479753.5 (0.66)51.8 (0.92)53.9 (1.09)61.6 (2.7))
    Race/ethnicity
        %White-non Hispanic471081.2 (1.20)74.8 (1.47)87.0 (1.13)86.4 (2.1)
        % Black non-Hispanic20848.9 (0.54)12.2 (0.73)5.6 (0.39)7.9 (0.90)
        % Mexican American20423.5 (0.26)5.2 (0.36)2.1 (0.19)1.2 (0.20)
        % Other3486.5 (0.91)7.8 (1.1)5.4 (0.96)4.5 (1.9)
Clinical condition (% present)
    History of ischaemic heart disease91799.2 (0.44)7.3 (0.62)9.4 (0.61)20.3 (2.0)
    History of myocardial infarction90856.2 (0.39)4.2 (0.45)6.3 (0.46)17.6 (1.80)
    History of angina66916.1 (0.45)5.3 (0.49)6.1 (0.69)10.3 (1.6)
    Chronic bronchitis91837.9 (0.39)8.4 (0.50)7.1 (0.51)10.2 (1.32)
    Congestive heart failure91633.7 (0.23)2.1 (0.28)3.6 (0.40)14.4 (1.80)
    Stroke91793.3 (0.25)2.2 (0.35)3.3 (0.32)10.9 (1.40)
    Diabetes (by history)91748.2 (0.37)7.0 (0.57)7.8 (0.41)18.0 (1.61)
    Hypertension (by history)914134.2 (0.83)29.6 (1.22)34.0 (1.24)63.9 (2.04)
    JNC VII staging of hypertension
        Normal < 120/80 %594068.8 (0.89)68.7 (1.38)69.0 (1.22)67.7 (2.5)
        Pre-hypertension 120/80 %227025.2 (0.69)25.9 (1.19)24.8 (0.95)23.7 (2.02)
        Stage I 140/90 %5385.2 (0.37)5.0 (0.49)5.2 (0.68)6.6 (1.32)
        Stage II 160/100 %1140.80 (0.14)0.45 (0.10)0.97 (0.21)2.0 (0.74)
    Family history of myocardial infarction < 50 yrs899715.3 (0.62)16.9 (1.02)14.2 (0.90)11.9 (1.7)
    Tobacco use9184
        Current smokers %198522.9 (0.90)28.3 (1.23)19.0 (1.03)12.6 (1.58)
        Former smokers %304835.0 (0.79)32.2 (1.05)37.0 (1.46)41.1 (1.79)
        Never smokers %415142.1 (0.84)39.5 (1.44)44.0 (1.30)46.3 (2.20)
    Physically inactive (%)918417.9 (0.82)18.4 (0.91)15.2 (1.04)31.5 (2.30)
    Body mass index (kg/m2)915227.3 (0.12)26.1 (0.14)26.8 (0.13)27.8 (0.20)
Laboratory variables
    Plasma fibrinogen (µmol/l)91849.0 (0.08)8.8 (0.08)9.0 (0.08)10.4 (0.19)
    Total cholesterol (mmol/l)91605.6 (0.02)5.5 (0.02)5.7 (0.03)6.0 (0.07)
    HDL cholesterol (mmol/l)90871.3 (0.01)1.3 (0.01)1.3 (0.01)1.3 (0.02)
    Total cholesterol/HDL ratio90844.7 (0.05)4.6 (0.07)4.7 (0.06)5.2 (0.1)
    C-reactive protein (nmol/l)91754.48 (0.01)4.29 (0.01)4.29 (0.01)6.95 (0.06)
    Ferritin (ng/ml)9160150.6 (2.81)151.3 (3.6)149.9 (4.0)149.9 (5.3)
    Serum creatinine (µmol/l)918497.7 (0.39)85.1 (0.21)103.1 (0.26)145.1 (2.6)
    eGFR MDRD (ml/min/1.73 m2)c918489.5 (0.51)107.7 (0.36)77.4 (0.21)48.7 (0.41)
    eGFR CKD-EPI ((ml/min/1.73 m2)d918487.5 (0.48)102.5 (0.30)78.6 (0.29)46.3 (0.45)
  • aMean values are reported with standard errors (±SE) in parenthesis. bConversion of µmol/l to mg/dl, multiply by 34.0. cGlomerular filtration rate (ml/min per 1.73 m2) was based on the abbreviated MDRD Study equation.24 dEstimated glomerular filtration rate (ml/min per 1.73 m2) was based on the CKD-EPI equation.25

Plasma fibrinogen and total mortality

During 10 years of follow-up, 2173 (15.7 %) died and 1068 were ascribed to cardiovascular disease. The crude all-cause mortality rate increased from 11.3 per 1000 person-years to 34.3 person-years from the lowest to the highest quartiles (Table 3). For each 1 µmol/l rise in plasma fibrinogen, overall mortality increased by 15 % (95% CI 1.12–1.18). With adjustment for demographic characteristics and Framingham factors, the association of plasma fibrinogen with mortality was attenuated but remained significant (HR = 1.07, 95% CI 1.04–1.09). Similarly, when plasma fibrinogen was modelled in categories, the adjusted hazard of death increased in a linear fashion from 1.25 (95% CI 1.03–1.52) in the second quartile to 1.51 (95% CI 1.25–1.83) in the fourth quartile.

View this table:
Table 3

Relationship of plasma fibrinogen (µmol/l) with total mortality and stratified by level of kidney function

ContinuousQuartiles
FirstSecondThirdFourth
Count (n)Per 1 µmol/l(0 < 7.7)(7.7 to <9.0)(9.0 to <10.5)(≥10.5)
(n = 1807)(n = 2152)(n = 2391)(n = 2834)
All participants
    Total alive (n, %)7011 (84.3)1512 (90.3)1759 (88.7)1873 (84.6)1867 (73.5)
    Total deaths (n, %)2173 (15.7)295 (9.7)393 (11.3)518 (15.4)967 (26.5)
    Unadjusted death rate per 1000 person-years11.313.619.334.3
HR for death and 95% CI
ModelReferent group
    Unadjusted1.15 (1.12–1.18)1.001.22 (0.99–1.50)1.91 (1.61–2.27)3.21 (2.64–3.90)
    Plus age1.09 (1.07–1.12)1.000.91 (0.74–1.14)1.23 (1.02–1.48)1.69 (1.41–2.02)
    Plus sex, race1.10 (1.07–1.12)1.000.94 (0.76–1.16)1.25 (1.04–1.51)1.76 (1.47–2.11)
    Plus Framingham risk factorsa1.07 (1.04–1.09)1.000.97 (0.78–1.21)1.25 (1.03–1.52)1.51 (1.25–1.83)
    Plus C-reactive proteinb1.03 (1.00–1.06)1.000.96 (0.77–1.19)1.21 (1.00–1.47)1.30 (1.05–1.62)
    Plus serum ferritinb1.03 (1.00–1.06)1.000.96 (0.77–1.20)1.21 (0.99–1.46)1.31 (1.05–1.62)
    Plus estimated glomerular filtration rate1.03 (1.00–1.06)1.000.95 (0.77–1.18)1.19 (0.98–1.45)1.27 (1.03–1.57)
Stratified by glomerular filtration rateContinuousQuartiles
FirstSecondThirdFourth
GFR > 90 ml/min4383Referent group
    Unadjusted1.15 (1.11–1.20)1.000.91 (0.59–1.41)1.53 (1.09–2.13)2.59 (1.83–3.66)
    Plus Framingham factorsa1.07 (1.03–1.11)1.000.81 (0.52–1.26)1.04 (0.74–1.48)1.45 (1.03–2.03)
GFR 60–90 ml/min
    Unadjusted35391.13 (1.08–1.17)1.001.20 (0.87–1.64)1.67 (1.19–2.35)2.44 (1.79–3.31)
    Plus Framingham factorsa1.05 (1.01–1.09)1.000.95 (0.68–1.32)1.16 (0.82–1.65)1.35 (1.00–1.83)
GFR < 60 ml/min903
    Unadjusted1.06 (1.01–1.11)1.001.48 (0.92–2.39)1.28 (0.79–2.06)2.17 (1.41–3.33)
    Plus Framingham factorsa1.06 (1.02–1.10)1.001.32 (0.83–2.09)1.12 (0.71–1.76)1.72 (1.14–2.58)
  • aFramingham factors include age (continuous), sex (male vs. female), diabetes, hypertension, blood pressure at examination, family history of heart disease, physical inactivity, smoking behaviour (never, previous or current smokers) and cholesterol/HDL ratio.

  • bC-reative protein and serum ferritin were sequentially added to the model as measures of inflammation.

Plasma fibrinogen and mortality by level of kidney function

Table 3 and Figure 1a–c illustrate the association of plasma fibrinogen with total mortality across categories of eGFR. The P-value for the interaction of fibrinogen with eGFR and mortality was not significant (P > 0.05). Among subjects with eGFR <60 ml/min/1.73 m2, each 1 µmol/l increase in plasma fibrinogen was associated with significantly elevated mortality risk in the adjusted analysis (HR = 1.06, 95% CI 1.02–1.10). The magnitude of association for plasma fibrinogen with mortality was virtually identical for subjects with eGFR >90 and eGFR 60–89 ml/min/1.73 m2 with HR = 1.07, 95% CI 1.03–1.11 and HR = 1.05, 95% CI 1.01–1.09, respectively.

Figure 1.

(a–c) Adjusted relationship between plasma fibrinogen and the hazard ratio of total mortality by eGFR. Note: Plasma fibrinogen was modelled by a restricted cubic spline with five knots at the, 5th, 25th, 50th, 75th and 95th percentiles corresponding to values of 6.1, 7.7, 9.0, 10.5 and 14.2 µmol/l using multivariable Cox regression.

Plasma fibrinogen and cardiovascular mortality

Cardiovascular mortality rates increased significantly from 4.6 to 16.0 per 1000 person-years with increasing quartile of fibrinogen as demonstrated in Table 4. In the unadjusted model, cardiovascular mortality risks increased with increasing quartile of plasma fibrinogen. With adjustment for demographic, Framingham and inflammatory factors, this trend of increasing cardiovascular mortality remained significant (P < 0.001) although the association was partially attenuated with adjustment for C-reactive protein.

View this table:
Table 4

Relationship of plasma fibrinogen (µmol/l) with cardiovascular mortality and stratified by level of kidney functiona

ContinuousPlasma fibrinogen quartiles
FirstSecondThirdFourth
Per 1 µmol/l(0 < 7.7)(7.7 to <9.0)(9.0 to <10.5)(≥10.5)
(n = 1807)(n = 2152)(n = 2391)(n = 2834)
All participantsSubjects (n)
    Total deaths (n, %)2173 (15.7)295 (9.7)393 (11.3)518 (15.4)967 (26.5)
    Cardiovascular deaths (n, %)1068 (7.3)142 (4.0)185 (5.1)263 (7.7)478 (12.3)
    Cardiovascular death rate per 1000 person-years4.66.19.716.0
HR death and 95% CI
ModelReferent group
    Unadjusted1.16 (1.12–1.19)1.001.52 (1.12–2.05)2.19 (1.64–2.93)3.90 (3.15–4.82)
    Plus age1.09 (1.06–1.12)1.001.08 (0.80–1.45)1.29 (0.98–1.71)1.83 (1.50–2.23)
    Plus sex and race1.09 (1.07–1.12)1.001.11 (0.82–1.49)1.32 (0.99–1.77)1.93 (1.57–2.36)
    Plus Framingham risk factorsb1.06 (1.03–1.09)1.001.13 (0.83–1.54)1.28 (0.95–1.72)1.61 (1.31–1.98)
    Plus C-reactive proteinc1.03 (0.99–1.06)1.001.12 (0.82–1.52)1.25 (0.93–1.67)1.43 (1.15–1.77)
    Plus serum ferritinc1.03 (0.99–1.06)1.001.12 (0.82–1.53)1.23 (0.92–1.63)1.43 (1.15–1.77)
    Plus estimated glomerular filtration rate1.02 (0.99–1.06)1.001.10 (0.81–1.50)1.21 (0.90–1.61)1.37 (1.10–1.71)
Stratified by level of kidney functionContinuousQuartiles
FirstSecondThirdFourth
GFR > 90 ml/min4383Referent group
    Unadjusted1.16 (1.09–1.23)1.000.90 (0.47–1.73)1.80 (0.95–3.44)2.68 (1.55–4.65)
    Plus Framingham factors1.06 (1.00–1.13)1.000.77 (0.41–1.47)1.12 (0.55–2.27)1.32 (0.78–2.24)
GFR 60–90 ml/min3539
    Unadjusted1.15 (1.11–1.20)1.001.27 (0.79–2.04)2.16 (1.39–3.35)2.91 (2.12–3.98)
    Plus Framingham factors1.06 (1.01–1.11)1.000.90 (0.56–1.47)1.37 (0.87–2.16)1.33 (0.96–1.84)
GFR < 60 ml/min903
    Unadjusted1.04 (0.99–1.09)1.001.55 (0.79–3.05)1.36 (0.71–2.60)2.03 (1.14–3.61)
    Plus Framingham factors1.03 (0.98–1.08)1.001.46 (0.78–2.74)1.26 (0.70–2.27)1.66 (0.94–2.93)
  • aGlomerular filtration rate (ml/min per 1.73 m2) was based on the MDRD Study equation. bFramingham factors include age (continuous), sex (male vs. female), diabetes, hypertension, family history of heart disease, physical inactivity, smoking behaviour (never, previous or current smokers) and cholesterol/HDL ratio.

  • cC-reative protein and serum ferritin were sequentially added to the model as measures of inflammation

Plasma fibrinogen and cardiovascular mortality by level of kidney function

The association of fibrinogen and cardiovascular mortality within eGFR categories is illustrated in Table 4 and Figure 2a–c. In general, these patterns were similar in magnitude to those described earlier between fibrinogen and all-cause mortality with the greatest risks for subjects in the highest quartile groups. For subjects, with GFR <60 ml/min, the positive trend of plasma fibrinogen with mortality persisted (HR = 1.03, 95% CI 0.98–1.08).

Figure 2.

(a–c) Adjusted relationship between plasma fibrinogen and the hazard ratio of cardiovascular mortality by eGFR. Note: Plasma fibrinogen was modelled by a restricted cubic spline with five knots at the, 5th, 25th, 50th, 75th and 95th percentiles corresponding to values of 6.1, 7.7, 9.0, 10.5 and 14.2 µmol/l using multivariable Cox regression.

Ancillary analysis

To further explore the robustness of our findings, we tested the association of plasma fibrinogen with mortality using the CKD-EPI equation as our measure of eGFR. For each analysis, the patterns of associations with mortality were similar.25 Second, we repeated the analysis with plasma fibrinogen modelled in deciles and a similar pattern with mortality was observed.

Discussion

This study demonstrates significant and independent relationships of plasma fibrinogen with total and cardiovascular mortality among individuals with and without kidney impairment in the general population. We demonstrated that each 1 µmol/l increase in plasma fibrinogen was associated with significantly elevated mortality risk. The relationships of plasma fibrinogen with total and cardiovascular mortality were present among those with moderate to severe kidney impairment and had effect estimates that were similar in magnitude to those with normal kidney function. Furthermore, the associations of fibrinogen with cardiovascular mortality were independent of Framingham-type risk factors and markers of systemic inflammation. Our findings suggest that plasma fibrinogen contributes to mortality in subjects with kidney impairment to the same extent as it does in the general population.

The excess risk of cardiovascular death among patients with CKD remains unexplained.6 It is suspected, but not proven, that much of this excess risk may be attributable to elevations in non-traditional risk factors that accumulate with declining kidney function.7–10 Prior studies have demonstrated elevations in plasma fibrinogen among patients who have reached end-stage kidney disease and treated with either peritoneal or haemodialysis.16,17,26 More recently, other investigators have reported elevations in plasma fibrinogen levels among subjects with CKD compared with those without and that these were independent of pre-exiting subclinical or clinical cardiovascular disease.3,4,27 In support of these findings, we also found substantial increases in plasma fibrinogen levels in subjects with CKD compared with those with normal kidney function (10.4 vs. 8.8 µmol/l) and demonstrated that mean concentrations were significantly higher for subjects with increasingly lower GFR levels. Furthermore, we found that subjects with CKD had plasma fibrinogen levels that were generally in the range of the highest quartile groups of fibrinogen for the general population.

Whether plasma fibrinogen can be truly considered a cardiovascular risk factor in subjects across the spectrum of CKD has not been confirmed. Despite evidence of elevated plasma levels in the setting of reduced kidney function, there is no conclusive evidence that higher levels are predictive of increased mortality among non-dialysis cohorts. Although Shlipak et al. found an association between plasma fibrinogen and CV mortality in older adults, these findings did not extend to those with CKD suggesting that this novel risk factor may not explain the high cardiovascular mortality among CKD patients.19 This study extends the evidence for plasma fibrinogen as an important contributor to cardiovascular events among patients with non-dialysis dependent CKD. First, we demonstrated that plasma fibrinogen was significantly and independently associated with total and cardiovascular mortality among subjects with CKD (eGFR < 60 ml/min/1.73 m2). Second, we found that the magnitude of risk was similar in magnitude across all levels of impaired kidney function suggesting that plasma fibrinogen behaves no differently in subjects with CKD than it does among subjects with normal GFR. Third, although plasma fibrinogen correlated with several Framingham risk factors and inflammatory markers, the association with mortality was independent of these measures.

The relative contribution of serum fibrinogen as a novel cardiovascular risk marker has received increased interest from the scientific community.15,28,29 A recent meta-analysis of over 154 000 participants from 31 prospective studies found moderately strong associations between plasma fibrinogen and the risk of major cardiovascular and non-vascular mortality.15 However, in almost all published studies, subjects with reduced kidney function or CKD were generally excluded and therefore the mortality risk associations of plasma fibrinogen in this high-risk population with varying levels of kidney impairment have until now remained largely unknown. The findings from the current nationally representative study would suggest that plasma fibrinogen carries the same weight as a potential risk factor in CKD as it does in the general population.

There are several mechanistic pathways through which plasma fibrinogen may exert its effect on total and cardiovascular mortality risk. Fibrinogen plays a vital role in inflammation, atherogenesis and thrombogenesis.27–30 Through its interaction with leucocytes, it facilitates cell recruitment and maturation, and has a key role in leukocyte adhesion to the vascular endothelium.30–33 Plasma fibrinogen also mediates the adhesion of platelets on endothelial cells. It induces proinflammatory changes in leukocytes which facilitate phagocytosis and antibody-mediated leukocyte toxicity. There is strong evidence that fibrinogen can initiate atherogenesis and contribute to the growth of plaques.34,35 Fibrinogen is also involved in the final common pathways of the coagulation and platelet aggregation.36 The enzymatic cleavage of fibrinogen into fibrin monomers by thrombin and their subsequent linkage to fibrin polymers is a critical step in thrombus formation.

We accept that the current observational study has some limitations including loss to follow-up although minimal, lack of information on other potentially important explanatory variables and errors in measurement of baseline variables. Nevertheless, this study has several major strengths which enhance its internal and external validity; the sample large size and the ability to test the hypothesis in several interest groups; the standardized methods of data collection; the relatively long follow-up, the large number of events; the outcome of mortality as the end point; and the ability to adjust sequentially for a large number of known mortality predictors that were captured in the baseline questionnaire.

In conclusion, we found that elevated plasma fibrinogen levels independently associated with total and cardiovascular mortality among individuals with varying levels of kidney impairment in the general population. The adverse impact on cardiovascular death was not accounted for by Framingham risk factors or measures of inflammation and the excess risk was similar in magnitude and direction for subjects with and without impaired kidney function as determined by eGFR. Strategies that target elevated plasma fibrinogen levels in CKD may lead to reduction in cardiovascular risk and major adverse outcomes.

Funding

Health Research Board (HRB) of Ireland: the funder did not contribute to the design, performance or interpretation of the results of this study.

Conflict of interest: None declared.

Acknowledgments

Dr Stack had full access to all of the data in the study and takes responsibility for the integrity of the data and data analysis. All collaborators participated in the study design, analysis, interpretation and preparation of the final manuscript. Ms Donigiewicz was a recipient of the 2011 Health Research Board (HRB) Summer Scholarship. Ms Jessica Jeong assisted with manuscript preparation. None of the authors have any relevant competing interests to disclose.

References

View Abstract