OUP user menu

Impaired thrombolysis: a novel cardiovascular risk factor in end-stage renal disease

Sumeet Sharma, Ken Farrington, Robert Kozarski, Christos Christopoulos, Maria Niespialowska-Steuden, Daniel Moffat, Diana A. Gorog
DOI: http://dx.doi.org/10.1093/eurheartj/ehs300 354-363 First published online: 9 October 2012


Aims End-stage renal disease (ESRD) patients have an excess cardiovascular risk, above that predicted by traditional risk factor models. Prothrombotic status may contribute to this increased risk. Global thrombotic status assessment, including measurement of occlusion time (OT) and thrombolytic status, may identify vulnerable patients. Our aim was to assess overall thrombotic status in ESRD and relate this to cardiovascular risk.

Methods and results Thrombotic and thrombolytic status of ESRD patients (n = 216) on haemodialysis was assessed using the Global Thrombosis Test. This novel, near-patient test measures the time required to form (OT) and time required to lyse (lysis time, LT) an occlusive platelet thrombus. Patients were followed-up for 276 ± 166 days for major adverse cardiovascular events (MACE, composite of cardiovascular death, non-fatal MI, or stroke). Peripheral arterial or arterio-venous fistula thrombosis was a secondary endpoint. Occlusion time was reduced (491 ± 177 vs. 378 ± 96 s, P < 0.001) and endogenous thrombolysis was impaired (LT median 1820 vs.1053 s, P < 0.001) in ESRD compared with normal subjects. LT≥3000 s occurred in 42% of ESRD patients, and none of the controls. Impaired endogenous thrombolysis (LT≥3000 s) was strongly associated MACE (HR = 4.25, 95% CI = 1.58–11.46, P = 0.004), non-fatal MI and stroke (HR = 14.28, 95% CI = 1.86–109.90, P = 0.01), and peripheral thrombosis (HR = 9.08, 95% CI = 2.08–39.75, P = 0.003). No association was found between OT and MACE.

Conclusion Impaired endogenous thrombolysis is a novel risk factor in ESRD, strongly associated with cardiovascular events.

  • Platelets
  • Thrombosis
  • End-stage renal disease
  • Haemodialysis
  • Cardiovascular risk


The risk of cardiac death in dialysis patients <45 years is 100-fold greater1 and the risk of non-fatal cardiovascular disease is 10–30 times higher in patients with end-stage renal disease (ESRD) compared with general population.2 Presence of angiographically significant coronary artery disease (CAD) ranges from 25% in young non-diabetic haemodialysis patients to 85% in older, diabetics with ESRD.3 Many risk factors are more prevalent in ESRD and may explain some, but not all, the increased cardiovascular risk,4 pointing towards the contributory role of non-traditional risk factors in ESRD, which traditional predictive models, such as the Framingham risk score, do not cover.

The identification of blood markers of inflammation, platelet hyper-reactivity, and hypercoagulability has received much attention to identify vulnerable patients.57 Thrombosis and thrombolysis are dynamic processes occurring simultaneously. The thrombotic–thrombolytic equilibrium may determine the clinical manifestation of an acute thrombotic event.8

Bleeding tendency associated with chronic kidney disease (CKD) is attributed to a deficiency in primary haemostasis. Paradoxically, ESRD has been termed a hypercoagulable state, associated with thrombosis.9 Enhanced platelet aggregation and elevation of thrombotic markers including thrombin–antithrombin III complex, D-dimer, and tumour necrosis factor has been demonstrated in dialysis patients1014 although dialysis may improve the thrombotic profile.12 Chronic kidney disease patients also express markers of impaired fibrinolysis such as increased fibrinogen, plasminogen activator inhibitor-1, and reduced tissue plasminogen activator.1518

Despite several studies demonstrating a significant relationship between non-responsiveness to antiplatelet medication and subsequent thrombotic events, great differences exist between the predictive powers of different platelet function tests. Those currently in clinical use measure the response of platelets to a specific agonist,19 despite the involvement of many other physiologically important factors such as high shear stress and thrombin generation. The first and predominant stimulus for platelet activation in a severely stenosed artery is the pathological high shear stress (>10 000 s−1) creating rapid and strong bonds between platelets without prior activation.20,21 Such shear-induced platelet activation brings about the release of soluble agonists (thromboxane, ADP) from circulating platelets and generation of thrombin.22 Extracellular matrix proteins mediate adhesion of platelets and primary platelet aggregates to the vessel wall.22 A major limitation of all point-of-care platelet function tests is the use of anticoagulated blood (citrate, heparin, hirudin, or other anticoagulant), which prevents the assessment of thrombin generation by activated platelets, the ultimate and major determinant of occlusive platelet-rich thrombus formation.19 Further, there is no test in current use to measure endogenous thrombolytic activity (lysis of a platelet-rich arterial thrombus) as opposed to fibrinolysis of a venous (red cells rich) thrombus. Since the measurement of individual components of the fibrinolytic pathway fails to give a realistic assessment of overall thrombolytic status, its value is questionable.

The Global Thrombosis Test (GTT) is the first clinically available, comprehensive, near-patient test to simultaneously measure thrombotic occlusion time (OT), coagulation, and spontaneous endogenous thrombolytic activity. As the test employs non-anticoagulated blood, it is different, and free from many of the shortcomings of conventional point-of-care platelet function assays that employ anticoagulated blood.23,24 We showed that in acute coronary syndrome (ACS) patients, impaired endogenous thrombolysis is highly predictive of cardiovascular events.25 The endogenous thrombolytic status of stable patients has not been well characterized, and may be particularly important in those at highest risk.

It was our aim to characterize thrombotic status in ESRD and investigate whether this novel test of overall thrombotic status could identify those who, despite renal replacement therapy, remain at risk of future cardiovascular events. We also included arterio-venous (AV) fistula thrombosis since this bears many similarities to arterial thrombosis. It occurs in the presence of arterial blood, under systolic pressure and pulsatile flow. In 80% of cases, it associated with a significant stenosis, at or close to the AV anastomosis, predominantly due to intimal hyperplasia and associated with inflammation.26,27 Abnormal haemodynamic shear stress is the most important upstream factor responsible for AV fistula failure. High shear rates upstream of a stenosed AV fistula predispose to platelet thrombus formation (rather than erythrocyte- and fibrin-rich thrombi formed in low shear, venous settings) and this is supported by histological findings.2731


Study population

Patients with ESRD aged 18–90 years established on haemodialysis for at least 3 months (n = 216) were tested after obtaining written informed consent. Subjects were excluded for the following reasons: known haematological disorder, bleeding diasthesis, blood dyscrasia (platelets <70, Hb <8 g/dL, INR>1.4, APTT>×2 UNL, leucocyte count <3.5 × 109/L, neutrophil count <1 × 109/L), warfarin or other anticoagulant treatment, thrombolysis or glycoprotein IIb/IIIa inhibitor prior to sampling, intercurrent illness, such as pneumonia, sepsis, ACS or heart failure in last 3 months; malignancy or other disease shortening life-expectancy to <12 months; participation in another study; inability to give informed consent or when follow-up over 1 year period was unlikely. A group of 10 ESRD patients receiving peritoneal dialysis (not taking antiplatelet medication) were also tested immediately before dialysis. Healthy volunteers (n = 100, age 38 ± 11 years) not taking medication, matched proportionally for sex and race, were also tested. Subjects were recruited, through advertisement, from among hospital staff and from relatives and carers of patients attending the outpatient department. The local research Ethics Committee approved the study.

Blood collection

From each subject, 9 mL venous blood was obtained from a peripheral (non-fistula) vein and tested immediately after withdrawal. In ESRD patients, blood was obtained immediately prior to dialysis, before the routine low-molecular weight heparin was administered according to local haemodialysis protocol. Low-molecular weight heparin is used in the majority of renal centres in the UK, as a single intravenous bolus provides predictable anticoagulation, while use of unfractionated heparin requires repeated boluses or continuous infusion. Blood was taken using an 18-gauge butterfly cannula using a two-syringe technique. The first 6 mL was used for routine tests (blood count, biochemistry, bicarbonate, and bone-profile) and the next 3 mL for global thrombotic status assessment. Blood is aspirated into a standard polypropylene syringe, which is directly, and immediately inserted into the fitting in the GTT instrument (within <15 s of withdrawal), that is positioned next to the subject to minimize sampling delay. The measurement is automated and starts as soon as blood is introduced. To ensure that the low-molecular weight heparin received just prior to the last dialysis session did not interfere or affect our results, we sampled 10 patients immediately before dialysis, exactly 48 h after the last dose of low-molecular weight heparin.

Assessment of thrombotic and thrombolytic status

Global thrombosis test

Thrombotic status and endogenous thrombolytic activity were assessed using the GTT (Montrose Diagnostics Ltd, UK). This is a novel, point-of-care assay that employs non-anticoagulated blood. The instrument measures the time taken to create shear-induced thrombi under pathophysiological conditions (discussed later) and in the second phase of the test, measures the time to achieve spontaneous lysis of thrombi created during the first phase. The principle of GTT has previously been described in detail25,32 and is shown in Figure 1. Blood is introduced into, and flows through, a plastic tube in which two metal balls are inserted in the conical part of the tube. There are four narrow gaps between the inner plastic surface and the balls. When blood flows through these gaps adjacent to the upper ball, the resulting high initial shear stress (180 dynes/cm2) causes activation of platelets. In the space between the balls, due to the turbulent flow and low shear, the activated platelets aggregate. Thrombin is generated, which accelerates the formation of these aggregates and stabilizes them through fibrin. When these stable thrombi reach the gaps around the lower ball, they gradually occlude these gaps, reducing the flow rate and finally arresting flow. The instrument measures the time (d) between consecutive blood drops, which increases gradually as flow slows down and at an arbitrary point (d≥15 s) the endpoint of the measurement is displayed (OT; seconds). The principle of the technique is shown in Figure 2. The restart of blood flow following occlusion is due to spontaneous thrombolysis (lysis time, LT; seconds). If lysis does not occur until 6000 s following OT (LT cut-off time), ‘no lysis’ is recorded.

Figure 1

Schematic showing principle of the Global Thrombosis Test (see text).

Figure 2

Distribution of (A) occlusion time and (B) lysis time of healthy volunteers and (C) occlusion time and (D) lysis time in end-stage renal disease. Y-axis shows number of subjects.

The GTT assesses thrombus formation under pathophysiological conditions, since (i) blood is non-anticoagulated, with physiological calcium-ion concentration, (ii) akin to pathological conditions, platelet-rich thrombus formation is initiated by high shear forces, with release of soluble agonists playing only a secondary role, and (iii) thrombin generation from shear-activated platelets (procoagulant activity of platelets) plays a major role in the formation and stabilization of thrombi.

Coefficient of variation was assessed by testing 10 healthy volunteers twice, at 48 h intervals and also by testing 10 ESRD patients twice at 48 h intervals.

Data collection and follow-up

Baseline demographics, including dialysis vintage, were collected along with urea kinetic adequacy parameters relating to the previous month's routine testing. Follow-up was performed at 30 days and then at 3-monthly intervals, and documents of all adverse events obtained. To assess the effect of haemodialysis on thrombotic status, a subgroup of 20 patients was tested immediately pre- and 2 h after dialysis. We assessed the effect of antiplatelet therapy on thrombotic status, in 20 healthy volunteers. Volunteers were tested before and 12 h after a loading dose of 300 mg aspirin, and after a month washout, assessed again before and 12 h after a loading dose of 300 mg clopidogrel.

Study endpoints

The primary endpoint of the study was the occurrence of major adverse cardiovascular events (MACE) defined as the composite of cardiovascular death, non-fatal myocardial infarction (MI), or cerebrovascular event. The secondary endpoint was the occurrence of peripheral vascular thrombosis including acute ischaemic limb and arterio-venous (AV) fistula thrombosis. Collection and adjudication of events was performed blinded to GTT results.

Cardiovascular events

New cardiovascular events were diagnosed in the presence of (i) cardiovascular death, defined as death from MI based on the Universal Definition of Myocardial Infarction33 (defined as sudden, unexpected cardiac death, involving cardiac arrest, often with symptoms suggestive of myocardial ischaemia, and accompanied by presumably new ST elevation, or new left bundle branch block, and/or evidence of fresh thrombus by coronary angiography and/or at autopsy, but death occurring before blood samples could be obtained, or at a time before the appearance of cardiac biomarkers in the blood), significant arrhythmia, or refractory congestive heart failure, or death attributed to cardiovascular cause at post-mortem; confirmed from death certificates as well as medical records and observers’ accounts; sudden death was included as a cardiovascular event, or (ii) non-fatal MI, defined according to the Universal Definition of Myocardial Infarction33 as a rise and/or fall of cardiac troponin with at least one value above the 99th percentile of the upper reference limit together with evidence of myocardial ischaemia with at least one of the following: symptoms of ischaemia; ECG changes indicative of new ischaemia (new ST-T changes or new left bundle branch block); development of pathological Q waves in the ECG; or imaging evidence of new loss of viable myocardium or new regional wall motion abnormality.

Cerebrovascular events

New-onset cerebrovascular event was suspected with recent onset of neurological symptoms or signs, e.g. aphasia, focal deficits, or unilateral paresis, thought to be vascular in origin and confirmed by computerized tomography or MRI. This included events with and without spontaneous clinical resolution, and thus included both stroke and transient ischaemic events.

Peripheral vascular events

Peripheral thrombotic events were defined as occurrence of acute ischaemic limb(s) or AV fistula thrombosis on clinical examination and confirmed by contrast angiography and/or duplex ultrasound.

Cause of death

Deaths were classified based on data obtained from post-mortem, from hospital notes, or from the general practitioner, regarding the patient's last illness, according to WHO criteria based on ICD 10 (2010).

Statistical analysis

A normal range for OT and LT was established from healthy volunteers. The required sample size was calculated based on the Cox PH prognostic model. On the basis of our earlier work,25 the unadjusted HR for LT dichotomized by 3000 s was 2.25 (95% CI: 1.34–4.7). Based on the assumption that we would see around 14% MACE events per year in the ESRD cohort,34,35 a sample size of 200 was predicted to yield 30 events/year and the final Cox PH model with four covariates would provide 5% significance and 80% power.36,37

Given 200 patients in the intervention group and arbitrary selected sample size of 100 healthy volunteers, the difference in LT (OT) between these groups is classified on the moderate (semi-large) effect size level. The Cox PH model was used to assess sample size and Mann–Whitney U test used to compare patients and healthy volunteers, with 5% significance and >90% power.

Unpaired t-test and Mann–Whitney U test were used for normally and non-normally distributed variables, respectively. Two-sided tests were used. Dichotomous variables were compared by χ2 test or Fisher's exact test, as appropriate. The univariate linear regression model was used to assess the relationship between a continuous variable (dependent) given a dichotomized variable (independent). Where necessary, log transformations were applied. Correlations were analysed using Spearman's rank test. Ability of OT or LT to discriminate between patients with and without MACE was evaluated by receiver operating characteristic (ROC) curve analysis.38 The optimal cut-off was determined by the value providing the greatest sum of sensitivity and specificity. Kaplan–Meier estimates with log rank tests were used to compare survival curves. Univariate Cox proportional hazard regression was performed on LT divided up into bands of 1000 s to investigate the relationship between LT and MACE, and to identify risk factors from which a multivariate Cox proportional hazard prognostic model was proposed. The hazard proportionality assumption was evaluated in the Cox model with scaled Schoenfeld residuals. The test was carried out for the univariate model including each of the candidate variables, and its multivariate versions: baseline (age, sex, haematocrit) and extended (age, sex, haematocrit, LT≥3000). In both setups, the hazard proportionality assumption was not rejected at 0.05 significance.

To assess the added predictive ability of LT≥3000 s for MACE, net reclassification improvement analysis39 was performed. For chosen risk cut-offs, models do not recognize patients within the low-risk group (<5%), among patients with and without events. The effect of interventions (heparin, aspirin, clopidogrel, dialysis) on thrombotic status was evaluated with Wilcoxon signed rank test. All tests were two-sided and significance was defined as P< 0.05. Analyses were performed with R (R Foundation for Statistical Computing, Vienna, Austria).


The characteristics of study population are shown in Table 1. Follow-up was available in all patients for 276 ± 166 days. The coefficient of variation for OT and LT was 8 and 10% in normal volunteers, and 6 and 5% in ESRD, similar to earlier studies using this technique.32,40,41 There was no correlation between OT and LT (Spearman's, P = 0.15).

View this table:
Table 1

Baseline characteristics of end-stage renal disease patients and relationship to occlusion time and lysis time

OTOverall (n = 216)LT < 3000 (n = 126)LT ≥ 3000 (n = 90)P-value
Intercept (P-value)Slope (P-value)
Risk factors, n (%)
 Age (years)555.4 (<0.001)−1.0 (0.218)64.4 ± 14.865.562.90.169
 Male gender459.8 (<0.001)48.4 (0.053)138 (63.9)80 (63.5)58 (64.4)1.000
 History of peripheral vascular disease492.2 (<0.001)−18.9 (0.672)17 (7.9)11 (8.7)6 (6.7)0.765
 Hypertension512.1 (<0.001)−30.7 (0.242)151 (69.9)88 (69.863 (70.0)1.000
 Hyperlipidaemia496.6 (<0.001)−8.0 (0.793)43 (19.9)29 (23.0)14 (15.5)0.143
 Diabetes478.6 (<0.001)42.2 (0.113)62 (28.7)35 (27.8)27 (30.0)0.839
 Prior CAD479.0 (<0.001)40.8 (0.125)62 (28.7)29 (23.0)33 (36.7)0.042
 Prior stroke491.6 (<0.001)−5.6 (0.862)37 (17.1)20 (15.9)17 (18.9)0.691
 Prior MI484.8 (<0.001)50.8 (0.177)25 (11.6)10 (7.9)15 (16.7)0.078
 LV dysfunction487.3 (<0.001)39.9 (0.255)32/153 (21.1)15/80 (18.8)17/73 (23.6)0.624
 Body mass index537.7 (<0.001)−1.8 (0.381)26 ± 626260.736
Drugs, n (%)
 Aspirin489.9 (<0.01)1.6 (0.947)111 (51.4)58 (46.0)53 (58.9)0.084
 Clopidogrel481.5 (<0.001)82.5 (0.0306)24 (11.1)12 (9.5)12 (13.3)0.510
 ACE-inhibitor488.0 (<0.001)7.9 (0.758)73 (33.8)36 (28.6)37 (41.1)0.076
 Statins476.5 (<0.001)29.1 (0.226)105 (48.6)60 (47.6)45 (50.0)0.836
 β-blocker493.5 (<0.001)−8.0 (0.752)76 (35.2)45 (35.7)31 (34.4)0.962
 Calcium channel blocker493.0 (<0.001)−7.7 (0.768)65 (30.1)38 (30.2)27 (30.0)1.000
 Proton pump inhibitor484.9 (<0.001)10.7 (0.659)118 (54.6)56 (44.4)62 (68.9)0.001
 Diuretic485.1 (<0.001)13.5 (0.582)89 (41.2)58 (46.0)31 (34.4)0.117
 Insulin484.1 (<0.001)39.3 (0.224)36 (16.7)17 (13.5)19 (21.1)0.195
 Erythropoietin427.7 (<0.001)67.0 (0.185)203 (94.0)120 (95.2)83 (92.2)0.530
Laboratory results
 Haemoglobin (g/dL)422.5 (<0.001)6.2 (0.542)11.0 ±
 Haematocrit500.1 (<0.001)−0.3 (0.934)32.8 ± 3.533.032.40.178
 Platelet count (×109)564.5 (<0.001)−0.3 (0.0419)227 ± 75.4228.9225.90.818
 Urea (mmol/L)377.1 (<0.001)5.6 (0.005)20.5 ± 6.820.620.40.894
 Creatinine (µmol/L)437.0 (<0.001)0.1 (0.128)732.1 ± 250722.2745.90.670
 Sodium (mmol/L)−804.3 (0.005)9.5 (0.005)136.5 ± 3.4136.3136.60.452
 Potassium (mmol/L)496.3 (<0.001)−1.1 (0.944)5.0 ±
 Calcium (mmol/L)820.8 (<0.001)−136.5 (0.069)2.4 ±
 Phosphate (mmol/L)489.8 (<0.001)1.9 (0.961)1.6 ±
 C-reactive protein (mg/L)502.0 (<0.001)−19.0 (0.438)18.0 ± 23.716.220.60.050
 Albumin (g/L)458.6 (<0.001)0.9 (0.746)36.0 ± 4.436.335.60.120
 Bicarbonate (mmol/L)541.9 (<0.001)−2.2 (0.576)23.6 ± 3.123.423.80.347
 Cholesterol (mmol/L)561.5 (<0.001)−16.4 (0.121)4.1 ±
 LDL/HDL ratio569.3 (<0.001)−13.7 (0.302)3.5 ±
 Fibrinogen (g/L)534.5 (<0.001)−4.3 (0.527)5.7 ±
 Parathormone (pg/mL)492.1 (<0.001)−0.03 (0.899)52.3 ± 5157.1490.559
Dialysis parameters
 Duration of HD (min)581.0 (<0.001)−0.5 (0.164)187.9 ± 34.9185.3191.50.191
 Kt/V449.0 (<0.001)30.3 (0.447)1.3 ±
 Dialysis vintage (months)580.9 (<0.001)−0.5 (0.164)6.3 ±
 KRU (mL/min/m2)494.3 (<0.001)−2.3 (0.744)1.4 ±
 Fistula (natural)14289530.099
 Fistula (graft)3120.768
  • NS, non-significant; CAD, coronary artery disease; MI, myocardial infarction; LV, left ventricle; HD, haemodialysis; KRU, renal urea clearance; Dialysis vintage, time from start of dialysis to present (in months). Kt/V, number used to quantify dialysis treatment adequacy, where K is the dialyzer clearance of urea, t the dialysis time, and V the volume of distribution of urea, approximately equal to patient's total body water.

In ESRD, both OT (491 ± 177 vs.378 ± 96 s, P < 0.001) and LT were prolonged (median LT 1820 vs.1053 s, P < 0.001) compared with healthy volunteers (Figure 2). None of the controls, but 41.7% ESRD patients had LT≥3000 s, with 34% demonstrating markedly impaired thrombolytic status with LT≥6000 s. There were 12 non-cardiovascular deaths, attributable to sepsis (2), pneumonia (1), ESRD (4), haematemesis (1), gastrointestinal haemorrhage, unspecified (2), enterocolitis due to Clostridium difficile (1), and malignant neoplasm colon (1). In the remainder, 23 MACE and 17 peripheral thrombotic events occurred [15 AV-fistula thromboses (14 in natural fistulae, 1 in an AV graft) and two acute ischaemic limbs] (Table 2 and Figure 3A).

View this table:
Table 2

Breakdown of major adverse cardiovascular events and hazard ratio based on lysis time

Overall (n = 216)LT < 3000 (n = 126)LT ≥ 3000 (n = 90)HR (CI)P-valueHR (CI) Adjusted for sex, age, HctP-value
Cardiovascular death, non-fatal MI, cerebrovascular accident23 (10.6)5 (4.0)18 (20)4.25 (1.58–11.46)0.0044.37 (1.58–12.12)0.005
Non-fatal MI and cerebrovascular accident13 (6.0)1 (0.8)12 (13.3)14.28 (1.86–109.90)0.01115.76 (2.00–124.06)0.009
Cardiovascular death10 (4.6)4 (3.2)6 (6.7)1.75 (0.49–6.22)0.3851.64 (0.44–6.15)0.465
Non-fatal MI9 (4.2)0 (0.0)9 (10.0)
Cerebrovascular accident4 (1.9)1 (0.8)3 (3.3)3.57 (0.37–34.42)0.2704.89 (0.46–51.4)0.186
Peripheral thrombotic event17 (7.9)2 (1.6)15 (16.7)9.08 (2.08–39.75)0.00310.81 (2.45–47.68)0.002
  • Values are given as n (%).

  • CI, confidence intervals; Hct, haematocrit; MI, non-fatal myocardial infarction; peripheral thrombotic events, composite of fistula thrombosis and acute ischaemic limb.

Figure 3

(A) Incidence of study endpoint events based on lysis time. MI, myocardial infarction, *P < 0.05. (B) Hazard ratios (HR) for major adverse cardiovascular events by lysis time. LT ≥ 3000 s was the optimal cut-point to predict major adverse cardiovascular events. The 95% confidence interval is shown in brackets. *P < 0.001; **P < 0.0001.

Survival analysis demonstrated a strong relationship between LT and MACE (vide infra), but not between OT and MACE (HR = 1, 95% CI = 0.9976–1.002, P = 0.969). The optimal cut-point correlating with MACE was LT≥2940 s (rounded to 3000 s for clinical ease) (Figure 4A and B).

Figure 4

Receiver operating curves for (A) occlusion time and (B) lysis time and (C) receiver operating curve analysis showing improvement of the prognostic model by adding LT ≥ 3000 s. Lysis time significantly discriminated between patients with and without major adverse cardiovascular events. AUC, area under the curve.

LT≥3000 s was associated with a significantly higher risk of MACE overall (HR = 4.25, 95% CI = 1.57–11.46, P = 0.004), non-fatal MI or cerebrovascular events (HR = 14.28, 95% CI = 1.86–109.9, P = 0.01), and peripheral thrombotic events (HR = 9.08, 95% CI = 2.08–39.75, P = 0.003) (Table 2). All non-fatal MIs and 15 of 17 peripheral thrombotic events occurred in those with LT≥3000 s. Non-cardiac death was not related to LT. As LT increased, HR increased up to LT≥3000 s (Figure 2B). There was no additional risk beyond 4000 s, possibly because a large number of subjects had severely impaired LT (LT≥6000 s, Figure 1). Impaired endogenous thrombolytic status was a very strongly associated with MACE (Figure 5).

Figure 5

Kaplan–Meier curves showing probability of event-free survival in end-stage renal disease based on lysis time. Thrombolytic status was strongly predictive of major adverse cardiovascular events. LT ≥ 3000 s was associated with HR 4.25 compared with those with LT < 3000 s.

All patient characteristics and variables were interrogated for effects on OT and LT, and shown in Table 2. Patients on clopidogrel exhibited longer (less thrombotic) OT than those not taking clopidogrel (564 ± 179 vs. 482 ± 175 s, P = 0.021). Occlusion time was also directly correlated with serum sodium and urea level. LT≥3000 s was significantly associated with raised fibrinogen levels, history of CAD, and proton pump inhibitor use.

A univariate model showed that only the following variables from Table 2 were related to MACE: serum calcium (P = 0.038), (log) CRP (P = 0.04), haematocrit (P = 0.013), and a negative relation with diuretic treatment (P = 0.044).

After a back-step model selection procedure, the following three variables were then entered into the baseline multivariate Cox proportional hazard model: haematocrit (HR = 0.87, 95% CI = 0.78–0.97, P = 0.015), and two traditional risk factors, namely age (HR = 0.999, 95% CI = 0.97–1.03, P = 0.944) and male sex (HR = 1.24, 95% CI = 0.5–3.03, P = 0.645).

None of these basic covariates were correlated either with LT or its dichotomized version, which would have increased the standard error of the hazard ratio in the Cox proportional hazard model. Multivariate Cox proportional hazard analysis including the baseline covariates showed that LT≥3000 s remained strongly associated with MACE after adjustment for the baseline risk factors (HR = 4.37, 95% CI = 1.58–12.12, P = 0.005). The baseline model (age, sex, haematocrit) was then extended by including LT≥3000 s to give the final predictive model. Of the baseline model covariates, only haematocrit remained significant in the extended model adjusted for age, sex, and LT (HR = 0.89, 95% CI 0.79–0.99).

Receiver operating characteristic curve analysis (Figure 3C) indicates an improvement in the area under the curve with the extended model with borderline significance (DeLong test Likelihood ratio test P = 0.058) indicating the usefulness of net reclassification improvement analysis, as ROC may not be sensitive enough to detect the new marker improvement.39 Reclassification analysis performed on the aforementioned baseline model with respect to the extended model (which included LT) showed that adding LT to the baseline risk factor model improves risk stratification of patients with ESRD in terms of cardiovascular risk. Inclusion of LT≥3000 s in the model containing three baseline predictors (haematocrit, age, sex) significantly added to the model effectiveness (net reclassification improvement = 0.61, P < 0.001) leading to improvement in reclassification mainly of non-event patients (see Supplementary material online, Figure S1 and accompanying Table).

Thrombotic status was not affected by haemodialysis (pre- and post-dialysis OT: 570 ± 138 vs. 545 ± 126 s, P = 0.368 and LT 1725 vs. 1665 s, P = 0.753, respectively) nor by the low-molecular weight heparin administered 48 h earlier [anti-FXa level undetectable (0.0 iu/mL) in all samples]. In patients receiving peritoneal dialysis, thrombotic status did not differ significantly from patients on HD (OT 419 ± 134 vs. 491 ± 177, P = 0.672 and LT 2561 vs. 1820 s, P = 0.574). In volunteers, aspirin significantly prolonged OT (356 ± 54 vs.530 ± 99 s, P < 0.0001), but did not affect LT (1043 vs. 1049 s, P = 0.741). Clopidogrel also increased OT (365 ± 54 vs. 569 ± 84 s, P < 0.001) but did not affect LT (1043 vs.1067 s, P = 0.731). Occlusion time prolongation in response to clopidogrel was marginally more than in response to aspirin (P = 0.02).


Our main findings are (i) that ESRD patients have markedly impaired endogenous thrombolysis, compared with healthy volunteers and (ii) that such impaired thrombolysis is associated with a high risk of cardiovascular and peripheral thrombotic events.

The prolonged OT in ESRD patients suggests either impaired primary haemostasis, as reported earlier, or reflects the fact that half the patients were on at least one antiplatelet agent, or a combination of these factors. In a large meta-analysis examining the efficacy of antiplatelet agents in preserving dialysis access patency, antiplatelet agents appeared effective in reducing thrombosis in central venous catheters and AV shunts, but not in preventing AV graft thrombosis.42 OT was not predictive of MACE. This was similar to our findings in ACS patients on dual antiplatelet medication.25

Despite the relatively high prevalence of cardiovascular risk factors in ESRD,4 the Framingham risk score projected cardiovascular risk in ESRD is similar or somewhat higher than reference populations from the Framingham cohort or from the National Health and Nutrition Examination Survey (NHANES) III.43 Even studies predicting higher risk may underestimate the incidence of cardiovascular disease observed in dialysis patients or even transplant recipients.44 These point towards the contributory role of non-traditional risk factors in promoting cardiovascular risk in ESRD, which traditional predictive models do not cover. Although aggressive risk factor modification in CKD with statins,45 angiotensin-converting enzyme inhibitors,44 and normalization of haemoglobin with erythropoietin46 reduce cardiovascular events,47 thrombotic events continue to occur. Thus, the optimal management of cardiovascular risk in ESRD not only requires traditional risk factor modification, but also identification of newer, previously unknown, risk factors in this complex group, that may explain the excess cardiovascular events, which are higher than in any other disease state.

In this study, we show that impaired LT is strongly correlated with adverse cardiovascular events in ESRD, independent of other risk factors. This is predominantly attributable to an increased risk of MI in those with LT≥3000 s, although it is very unlikely that the LT<3000 s group is completely protected from MI. The MACE rate in those with LT<3000 s was much lower than anticipated and may have contributed to the lower than expected significance of our results. Although the hazard ratio for MACE is high, the sample size is relatively small and the confidence intervals large, thus our results could be due to chance. Although studies cannot be directly compared, our findings are supported by the prior study in ACS patients, where the same cut-off value (LT≥3000 s) was correlated with increased cardiovascular risk25 but imparted a much greater risk in ESRD than in ACS patients (HR = 4.25 vs. 2.5) and LT was more frequently impaired and longer in ESRD than in ACS patients.25 The distribution of events was not even in our subgroups (Figure 3B), with a break in events between 3000 and 5000 s, as there were no events in LT 4000–5000 s group. The reduction in hazard when LT>4000 s is likely, at least in part, to reflect the uneven distribution of events, with small numbers of patients and even smaller numbers of events in these groups.

Our findings are supported not only by studies showing increased fibrinogen, plasminogen activator inhibitor-1, and reduced tissue plasminogen activator in ESRD,4651 but also recent data on fibrin clot properties. Both ESRD52,53 and thrombo-embolic coronary events5456 have been associated with the formation of dense fibrin clots, resistant to fibrinolysis. Using turbidometric plasma clot lysis, fibrin clot permeability and perfusion clot lysis assays, Undas et al.57 showed that ACS and CKD patients have higher plasminogen activator inhibitor-1 and tissue plasminogen activator levels and formed fibrin clots that were less porous and more resistant to fibrinolysis, than a control group of ACS patients with normal renal function. The fibrin clot in CKD patients exhibited smaller pore size, larger number of protofibrils per fibrin fibre, increased fibre size and clot mass. Our finding of an association between raised fibrinogen level and impaired endogenous thrombolysis is interesting. Based on epidemiological data, fibrinogen has been associated with increased cardiovascular and arterial thrombotic risk, but whether this relationship is causal is not established. Recently, high fibrinogen levels have been linked with resistance to thrombolytic therapy and adverse outcome in patients with ischaemic stroke.58 Furthermore, in a murine model, artificial increase in fibrinogen level directly promoted thrombosis and thrombolysis-resistance, via enhanced fibrin formation and stability.59 However, in a transgenic mouse model of hyperfibrinogenaemia, mice with high fibrinogen level did not demonstrate accelerated platelet thrombus formation in response to injury, compared with wild-type.60 Surprisingly, transgenic mice demonstrated suppression of thrombin generation in plasma and activation of the fibrinolytic system. Furthermore, genetic variations such as the γ’ splice variation in fibrinogen gene transcription result in more highly cross-linked and stable fibrin clots, with reduced pore size, that are more resistant to lysis.61 This variant is associated with an increased risk of thrombosis and MI, an effect that is independent of fibrinogen levels.61,62

To investigate further the relationship between fibrinogen level and impaired endogenous thrombolysis, we performed an in vitro experimental correction to raise fibrinogen level and assess the effect on LT. In healthy volunteers (n = 10), in parallel measurements, increase in plasma fibrinogen concentration in vitro by 1 g/L did not significantly alter LT compared with control samples [median LT 1081 (IQR 907–1300) s and 1297 [1208–1660] s respectively, P = 0.06]. Thus increasing fibrinogen concentration showed a trend towards enhanced rather than inhibited spontaneous thrombolytic activity.

Our in vitro findings in healthy blood support others claiming that hyperfibrinogenemia per se does not inhibit fibrinolysis.60 Thus, it is more likely that it is not just the elevated plasma fibrinogen concentration per se, but the quality of the fibrin clot architecture that determines risk.

It is likely that in ESRD, fibrinogen structure and function are also altered, making clots more resistant to lysis52 and this may be causally related to increased thrombotic risk.

Unfavourable clot properties were demonstrated in ESRD patients on haemodialysis, and although a small study, there was an association between mortality and reduced clot permeability and prolonged LT.53 This may explain the functional significance of the impaired thrombolytic state observed in ESRD. We did not seek to compare LT with plasma markers of fibrinolysis, since the value of fibrinolysis activity markers is very limited in aiding diagnosis and risk stratification in the individual patient.63

Prolonged LT was associated with prior CAD but it is difficult to say whether it is a causative phenomenon. Lysis time in ACS patients was not related to prior CAD.25 Proton pump inhibitor use was related to prolonged LT. Our study is too small to analyse whether this reflects a possible effect of proton pump inhibitors inhibiting the CYP2C19 isoenzyme, thereby reducing the ability of clopidogrel to inhibit platelet aggregation. That higher serum calcium concentration was associated with prolonged LT may be functionally important. Ionized calcium (Ca2+) holds together the fibrinogen binding receptor glycoprotein IIb/IIIa complex, is essential for agonist-induced conversion of the glycoprotein IIb/IIIa complex into the functional fibrinogen receptor, and is required for the binding of fibrinogen to its receptor, as well as for the coagulation cascade.19 C-reactive protein was related to LT suggests a relationship between inflammation and thrombosis. Inflammation biomarkers are strong predictors of MI or thrombotic stroke64 and modification of platelet function has been reported to modulate inflammatory mediators.65

Limitations of our study include that normal volunteers were not age-matched, patients were sampled only once and pre-dialysis, the fact that only CKD patients on haemodialysis were studied, and that most patients received erythropoietin, which is known to increase the number of circulating platelets, improves platelet function,66 and has varying effects on platelet reactivity and fibrinolysis.67 It has also been proposed that in patients treated with erythropoietin, increased activity of C-reactive protein, nitric-oxide, and thrombin-activatable fibrinolysis inhibitor leads to a fibrinolytic deficit with resultant increase in thrombosis.68 The effect of antiplatelet medication on LT is not fully established, but unlikely to be significant given the findings in patients with ACS25 and here in ESRD. It will be interesting to see if LT is shortened by the emerging novel thrombin inhibitors once these are licensed for ACS, since a significant portion of patients with ACS have prolonged LT and this is a risk factor for recurrent cardiovascular events.25

Our identification of impaired thrombolysis as a novel risk factor in ESRD may have important implications for screening and risk stratification. Since thrombotic status does not appear to be affected by haemodialysis, future studies are required to investigate medical therapies to improve endogenous thrombolysis, to see whether this may reduce the risk of cardiovascular events in these high-risk patients.


E&N Hertfordshire NHS Trust.

Conflict of interest: D.A.G. is related through family to a company director in Montrose Diagnostics Ltd, but has no financial involvement or equity interest in, and has received no financial assistance, support, or grant from the aforementioned company.


View Abstract