Differential propensity for major hemorrhagic events in patients with different types of arterial disease


Sefanja Achterberg, Department of Neurology, University Medical Center Utrecht, Room H02.128, PO Box 85500, 3508 GA Utrecht, the Netherlands.
Tel.: +31 88 75558350; fax: +31 30 2522782.
E-mail: s.achterberg-2@umcutrecht.nl


Summary. Aims: Atherosclerosis is the most frequent cause of coronary artery disease (CAD), cerebrovascular disease (CVD), and peripheral arterial obstructive disease (PAD). We previously found that patients with CVD or PAD had a two-fold higher risk of major hemorrhagic complications than patients with CAD. We investigated whether this difference was attributable to baseline risk factors or genetic variants involved in hemostasis. Methods and results: We included 2622 consecutive patients from a single university hospital who presented with non-disabling CAD, CVD, or PAD. All patients were followed for the occurrence of major hemorrhagic complications for a mean of 6.6 years. Major hemorrhagic events included intracranial hemorrhagic events, fatal hemorrhagic events, and any hemorrhagic complications requiring hospitalization, irrespective of interventions. Major hemorrhagic complications occurred in 122 patients (annual event rate of 0.77%). Patients with CVD or PAD had more hemorrhagic complications than patients with CAD (hazard ratio [HR] 2.05, 95% confidence interval [CI] 1.39–3.01). Hypertension, diabetes, renal failure and use of oral anticoagulants or antiplatelet therapy did not explain the difference (HR adjusted for all characteristics 1.74; 95% CI 1.14–2.61). Additional adjustment for genetic variants did not further change the HR. Conclusion: Patients with CVD or PAD are at higher risk for major hemorrhagic events than patients with CAD. This difference could not be explained by known risk factors, use of antithrombotic agents, or genetic variants involved in hemostasis. Further research to find the reason for this difference and possible differences in pathogenesis is warranted.


Patients with coronary artery disease (CAD), cerebrovascular disease (CVD) or peripheral arterial obstructive disease (PAD) are thought to represent different clinical expressions of atherosclerosis. Atherosclerotic plaque rupture, arterial thrombosis and thromboembolism can affect different arterial beds and, depending on arterial size and the severity of the distal occlusion, may lead to different clinical manifestations. Patients with atherosclerotic vascular diseases are at risk for new ischemic events as well as hemorrhagic events [1]. To prevent future cardiovascular ischemic events, treatment with antithrombotic agents, including anticoagulants or antiplatelet therapy, is recommended [2–4].

In a previous study, we found that patients with CVD of presumed atherosclerotic origin or with PAD had an increased risk of major hemorrhagic events as compared with patients with CAD [5]. This differential bleeding propensity may be attributable to a different distribution of risk factors for hemorrhage among the three presentations of vascular disease. Moreover, genetic variants involved in hemostasis could be distributed unequally between the three groups. Finally, preventive antithrombotic treatments might be different. Hence, we investigated whether this difference in hemorrhagic risk was associated with baseline risk factors, use of antithrombotic agents, or genetic variants involved in hemostasis.


Study design and patient population

Patients aged 18–79 years, newly referred to the University Medical Center Utrecht, The Netherlands, with classic risk factors for arterial disease (hypertension, hyperlipidemia, or diabetes mellitus) or with symptomatic arterial disease (CAD, CVD, PAD or abdominal aortic aneurysm), were included in the Second Manifestations of ARTerial disease (SMART) study (for definitions, see Table S1). A detailed description of the study was published previously [6]. Briefly, patients who gave written informed consent underwent a standardized vascular screening program, including a health questionnaire, laboratory assessment, and ultrasonography, to investigate the prevalence of additional vascular diseases. The Ethics Committee of the hospital approved the study.

For the current study, we used data only from patients with CAD, CVD or PAD, and we prespecified the comparison of CVD and PAD patients vs. CAD patients.

Genetic variant selection and genotyping

Twenty-two variants in 14 genes based on previous associations with ischemic stroke or myocardial infarction or functionality studies were selected in an earlier study and determined at that time because of involvement in atherothrombotic diseases [7]. We hypothesized, in accordance with the literature, that, without these genetic variants, the risk of hemorrhagic events might be higher [8].

DNA was isolated from blood samples and amplified by PCR. Genotyping was performed with the 5′-nuclease/TaqMan assay. PCRs with fluorescent allele-specific oligonucleotide probes (Assay-by-Design/Assay-on-Demand; Applied Biosystems, Foster City, CA, USA) were performed on a PTC-225 thermal cyler (Biozym, Hessisch Oldendorf, Germany), and fluorescence endpoint reading for allelic discrimination was performed on an ABI 7900 HT (Applied Biosystems). All genotypes were determined without knowledge of patient characteristics and outcome.


Primary outcome was defined as the first occurrence of a fatal or non-fatal hemorrhagic event. This included any intracranial hemorrhagic event, fatal hemorrhagic event, and any hemorrhagic complication requiring hospitalization, irrespective of intervention. For potential outcome events reported by the patient, we retrieved hospital discharge letters and the results of relevant laboratory and radiology examinations. Three members of the SMART Outcome Event Committee independently audited all events on the basis of available information. This committee consisted of physicians from different departments. In cases of disagreement, consensus was reached by consulting other members of the Outcome Event Committee.

Data analyses

The incidence of major hemorrhagic complications in patients with PAD, CVD or CAD was compared with Cox regression analysis, which generated hazard ratios (HRs) with accompanying 95% confidence intervals (CIs). Patients with incomplete follow-up were censored at the last time of observation. We used Kaplan–Meier curves for graphical display of the data. We compared patients with PAD or CVD together with patients with CAD, as there were no differences in hemorrhagic risk between patients with CAD and PAD. We first calculated crude HRs, and then assessed the influence of risk factors and genetic variants in bivariable analyses, yielding adjusted HRs. Variables changing the crude HR by more than 5% were included in a multivariable model. The influence of genetic variables was assessed in a similar way. For the analysis of genetic variants, we used a dominant model of inheritance: variant allele heterozygotes and homozygotes were compared with wild-type homozygotes.

In addition, we performed a sensitivity analysis, for which we selected only patients without clinically manifest disease in vascular beds other than the one with which the patient enrolled into SMART. For example, patients included with symptomatic coronary vessel disease with a history of peripheral obstructive arterial or cerebrovascular disease were excluded from these analyses.



For the current study, the data of 2622 consecutive patients presenting with transient or non-disabling manifestations of CAD (1237), CVD (711) or PAD (674) with information on genetic variants were available. These patients were included between September 1996 and April 2005, and were followed until March 2009 or death.

Baseline characteristics of risk factors for hemorrhagic complications are summarized in Table 1. The majority of patients were male, and the mean age was 60 years. Male sex, renal failure, hypertension and diabetes were more common in patients with CVD or PAD than in patients with CAD. At baseline, the use of oral anticoagulant drugs was not different between the three patient groups, and platelet inhibitors and statins were used less by patients with PAD than by the other patients.

Table 1.   Baseline characteristics of patients with cerebrovascular, peripheral arterial obstructive and coronary artery disease
 Cerebrovascular disease
(n = 711)
Peripheral arterial obstructive disease
(n = 674)
Coronary artery disease
(n = 1237)
  1. SD, standard deviation. *At time of inclusion. Mostly hemorrhage in the gastrointestinal tract.

Age (years), mean (SD)61.7 (10.4)59.1 (10.7)58.4 (9.4)
Male sex, no. (%)524 (73.7)448 (66.5)1013 (81.9)
History of stroke, no. (%)277 (38.9)40 (5.9)21 (1.7)
Hypertension*, no. (%)220 (30.9)197 (29.2)198 (16.0)
Diabetes, no. (%)116 (16.7)122 (18.7)152 (12.6)
Renal failure, no. (%)63 (9.5)84 (13.3)50 (4.3)
Platelet inhibitors*, no. (%)525 (73.8)274 (40.7)869 (70.3)
Oral anticoagulants*, no. (%)60 (8.4)66 (9.8)94 (7.6)
Statins*, no. (%)159 (22.4)89 (13.2)430 (34.8)
Number of hemorrhagic events during follow-up, no. (%)44 (6.2)41 (6.1)37 (3.0)
Annual risk of hemorrhagic events (%)0.910.940.45
Type of bleeding, no. (%)
 Intracranial hemorrhage, non-fatal5 (11.4)4 (9.8)3 (8.1
 Intracranial hemorrhage, fatal4 (9.1)3 (7.3)1 (2.7)
 Aortic aneurysm, non-fatal2 (4.6)00
 Aortic aneurysm, fatal2 (4.5)1 (2.4)0
 Extracranial hemorrhage: admission†19 (43.2)21 (51.2)26 (70.3)
 Extracranial hemorrhage: intervention-related12 (27.3)12 (29.3)7 (18.9)

Follow-up and occurrence of hemorrhagic events

The median follow-up times were 6.8 years in the CVD cohort, 6.6 years in the CAD cohort, and 6.4 years in the PAD cohort. Follow-up was complete for 98.1% of the patients. The annual risk of hemorrhagic events was 0.91% in the CVD cohort (44 events), 0.94% in the PAD cohort (41 events), and 0.45% in the CAD cohort (37 events) (Table 1). Figure 1 shows the unadjusted time-to-hemorrhage curves for the three disease categories.

Figure 1.

 Kaplan–Meier curves for three atherosclerotic disease categories in regard to hemorrhagic events. CAD, coronary artery disease; CVD, cerebrovascular disease; PAD, peripheral arterial obstructive disease.

Patients with CVD and PAD had more hemorrhagic events than patients with CAD (HR 2.05, 95% CI 1.39–3.01). Age, sex, hypertension and renal failure had the largest influence on this risk difference (Table 2). Simultaneous adjustment for these four risk factors lowered the HR to 1.65 (95% CI 1.10–2.48). Bleeding complications in patients with PAD and CVD were more often intracranial hemorrhages than in patients with CAD (16 events out of 85 [19%] hemorrhages vs. four events out of 37 [11%] hemorrhages). Most of the hemorrhagic events occurred in the gastrointestinal tract (Table 1). Patients with CVD or PAD with a hemorrhagic event less often used platelet inhibitors at the time of the event than patients with CAD (59% vs. 75%, relative risk [RR] 0.78; 95% CI 0.60–1.00). Three patients with CAD and five patients with PAD or CVD used combined platelet inhibitors at the time of the hemorrhagic event. There was no difference in the use of oral anticoagulants at that time.

Table 2.   Adjusted hazard ratios for hemorrhagic events according to different vascular diseases
 Peripheral artery vs. cerebrovascular diseasePeripheral and cerebrovascular vs. coronary artery disease
  1. F3, tissue factor; F5, coagulation factor V; GP1BA, glycoprotein 1 b-alpha. Data are no. (%) or hazard ratio (95% confidence interval). *Use at time of inclusion. †Adjustment for most influential variables; age, sex, hypertension, and renal failure. ‡Adjustment for age, sex, hypertension, renal failure, and four genetic variants.

Hemorrhagic event85122
Unadjusted0.99 (0.65–1.52)2.05 (1.39–3.01)
Adjusted for
 Age (years)0.85 (0.56–1.31)1.88 (1.27–2.77)
 Sex0.96 (0.62–1.46)2.11 (1.43–3.11)
 Hypertension0.99 (0.65–1.51)1.86 (1.26–2.75)
 Diabetes0.99 (0.65–1.51)2.13 (1.44–3.15)
 Renal failure1.07 (0.69–1.66)1.83 (1.23–2.72)
 Platelet inhibitors*1.00 (0.63–1.59)2.07 (1.40–3.06)
 Oral anticoagulants*0.99 (0.65–1.52)2.05 (1.39–3.02)
 Multiple†0.91 (0.59–1.43)1.65 (1.10–2.48)
 GP1BA Thr145Met0.99 (0.65–1.51)2.05 (1.39–3.01)
 F3 A(–603)G1.00 (0.65–1.53)2.05 (1.39–3.01)
 F5 Arg506Gln0.99 (0.65–1.51)2.05 (1.39–3.01)
 All three genetic variants1.00 (0.65–1.52)2.04 (1.39–3.00)
 Multiple‡0.90 (0.58–1.41)1.64 (1.09–2.46)

Influence of prothrombotic genetic variants on the occurrence of hemorrhagic events

The overall call rate for all 22 genetic variants was 98.9% (range 95.8–99.7). Genotype-specific risks and HRs of major hemorrhagic events are presented in Table 3. None of the 22 variants was found to be associated with the risk of major hemorrhagic events. Additional adjustment of the HR adjusted for the four vascular risk factors with the three genetic characteristics most strongly related to the incidence of major hemorrhages did not change the HR further (Table 2).

Table 3.   Genetic variants and hemorrhagic risk
Gene symbolVariantRisk of new hemorrhagic event per genotypeHazard ratio (95% CI)
  1. CI, confidence interval; F2, coagulation factor II; F3, tissue factor; F5, coagulation factor V; F13A1, coagulation factor XIII subunit A; F13B, coagulation factor XIII subunit B; FGA, fibrinogen alpha; FGB, fibrinogen beta; FGG, fibrinogen gamma; TPA, tissue plasminogen activator; VKORC1, vitamin K epoxide reductase complex-1; VWF, von Willebrand factor; GP1BA, glycoprotein 1 b-alpha; GP6, glycoprotein VI. Odds: Wild type vs. heterozygote and homozygote variant genotype. *Numbers represent patients without major hemorrhagic event/patients with major hemorrhagic event.

F2G20210A2414/12157/11/00.33 (0.05–2.36)
F3A(–603)G743/271221/74523/211.49 (0.97–2.28)
F5Arg506Gln2326/118145/41/00.52 (0.20–1.40)
F13A1Val34Leu146/10910/461390/660.73 (0.38–1.39)
F13A1Tyr204Phe2337/113143/84/11.27 (0.65–2.51)
F13A1Pro564Leu1544/73833/36107/131.09 (0.76–1.56)
F13BHis95Arg2064/100410/1815/31.00 (0.63–1.60)
FGAThr312Ala1264/641017/45206/120.90 (0.63–1.29)
FGBG(–854)A1747/86681/3159/51.00 (0.68–1.47)
FGBG(–455)A1591/73785/4491/21.15 (0.79–1.66)
FGGG7874A1292/68990/45197/80.84 (0.58–1.20)
FGGT9340C1213/631028/47244/130.91 (0.64–1.30)
FGGG5836A2305/111175/93/01.08 (0.55–2.13)
TPAC7351T1061/551141/53283/130.88 (0.62–1.26)
VKORC-1T2255C998/481108/55380/181.03 (0.72–1.48)
VWFThr789Ala1147/581016/49273/150.98 (0.69–1.39)
GP1BAT(–5)C1875/86562/3349/31.24 (0.84–1.83)
GP1BAThr145Met2142/99331/2215/11.46 (0.93–2.30)
GP6Ser219Pro1731/88674/3273/10.89 (0.60–1.33)
GP6Thr249Ala1700/85697/5390/20.96 (0.65–1.41)
GP6Gln317Leu1696/84699/3689/20.99 (0.68–1.46)
GP6His322Asn1775/8765/3363/21.01 (0.70–1.51)

Sensitivity analysis

When we restricted our analyses to the 2234 patients without other clinically manifest arterial disease apart from the referral diagnosis, the results changed slightly: HR for PAD vs. HR for CVD from 0.90 (0.58–1.41) to 0.71 (0.40–1.24), and HR for PAD and HR for CVD vs. HR for CAD from 1.64 (1.09–2.46) to 1.42 (0.91–2.21).


Patients with PAD and CVD are more prone to major hemorrhagic events than patients with CAD. This risk difference was not explained by vascular risk factors, and neither did a selection of prothrombotic genetic variants explain the difference.

To the best of our knowledge, this is the first study describing differences in hemorrhagic risk in patients with different manifestations of atherosclerotic vascular disease. Many studies have focused on reduction of new thrombotic events rather than on hemorrhagic risk [9–11]. A meta-analysis showed that treatment with antithrombotics, mainly aspirin, was safe and useful to prevent new events in patients with CAD, CVD, and PAD [12]. The absolute benefit of the treatment in terms of vascular event reduction or mortality substantially outweighed the absolute risk of major extracranial hemorrhage. No analyses were shown on potential differences in bleeding risk between patients with different locations of atherosclerotic vascular diseases. Data on individual studies included in this meta-analysis were shown in the appendix of this article, and allowed us to make such comparisons ourselves. We found that, in the population with previous myocardial infarction, three patients suffered from a hemorrhagic event in 23 098 person-years (annual event risk of 0.01%); patients with a previous stroke or transient ischemic attack had 80 hemorrhagic events in 27 872 person-years (annual event risk of 0.29%); and patients with intermittent claudication had 60 hemorrhagic events in 4492 person-years (annual event risk of 0.35%) [12]. The comparison between PAD/CVD and CAD patients gives an incidence rate ratio of 23 (95% CI 7–72). These findings support our own results.

With restriction of our analysis to patients with clinical manifestation in only one vascular bed, the HR for PAD and CVD vs. CAD lost its statistical significance; 1.42 (95% CI 0.91–2.21). From this analysis, we inferred, however, that the main result remained essentially the same, even though the HR decreased somewhat. The loss of statistical significance in the sensitivity analysis could be mainly attributable to the smaller number of patients in this analysis.

Strengths of the present study include the large sample size and direct comparison of three groups of patients with different locations of vascular disease being treated in the same hospital and evaluated and treated according to standardized protocols, both at presentation and during follow-up. The university hospital-based origin of our series could limit the generalizability of our study findings. For example, our patients are slightly younger than patients with atherosclerotic disease from the general population. However, we believe that the comparison between the three disease categories with respect to major hemorrhagic events in follow-up has not been distorted by this selection. Moreover, the follow-up results are in agreement with those found in a major meta-analysis [12]. The aim was to include patients consecutively; however, we have no register of patients missed for inclusion, or patients excluded for some reason, or those who did not give informed consent, so we cannot be completely certain whether our series was completely consecutive.

A limitation of our study is that we were only informed about the use of antithrombotic drugs and statins at baseline of the study. Classes of medications such as platelet inhibitors were noted, but no detailed information on the precise type of drug that the patient used was recorded. During follow-up, changes in medication usage were also not recorded. To determine whether major changes occurred in drug regimens over the years, we reassessed the information (mainly discharge letters) from the patients with major hemorrhagic events. At the time of the hemorrhagic event, no difference was found in the use of oral anticoagulant therapy between PAD/CVD and CAD patients (RR 1.41, 95% CI 0.71–2.83), but PAD/CVD patients has less use of platelet inhibitors than CAD patients (RR 0.78, 95% CI 0.60–1.00). Thus, the lower level of major hemorrhagic events in the CAD group was associated with more use of platelet inhibitors.

Statins were prescribed as protection against new ischemic events, but may lead to a higher risk of hemorrhagic intracerebral events [13]. Patients with the lowest hemorrhagic risk, CAD patients, used the most statins at baseline (35% vs. 19% of patients with PAD/CVD), so it is unlikely that statin use can explain the findings of our study. The genetic variants tested in our patients did not explain the differences in hemorrhagic events. A few reports discuss single polymorphisms as possible risk factors for major hemorrhage, with mostly conflicting results [14,15]. The gathering of different variants in genome-wide association studies might provide more results regarding this complex disease [15].

The reason for the difference in hemorrhagic risk between patients with PAD/CVD and those with CAD is not clear. We found no explanation for this difference in the literature, and thus can only speculate about the origin of our finding. Possibly, patients with PAD and CVD have more fragile vessels, which, together with atherosclerotic disease of these vessels, rupture easily, causing hemorrhagic events.

In past publications, all atherothrombotic diseases were often thought to arise from the same mechanism, namely atherosclerosis [16]. In a previous study, we showed differences in baseline risk factors and follow-up between CAD, CVD and PAD patients [5]. The difference in hemorrhagic risk shown in the current analyses should be taken into account in future research, as it might point to different underlying mechanisms for different manifestations of what seems to be a single disease. Treatment options might not be the same for different atherosclerotic diseases when this difference in hemorrhagic risk is taken into account.


S. Achterberg and A. Algra: designed the research, analyzed the data, and drafted the manuscript. All authors contributed to the collection and interpretation of the data and critical revision of the manuscript.


S. Achterberg is supported, in part, by a grant from the Netherlands Heart Foundation, grant No. 2005B031.

Disclosure of Conflict of Interests

The other authors state that they have no conflict of interest.


Members of the SMART study group: A. Algra, Y. van der Graaf, D. E. Grobbee, and G. E. H. M. Rutten, Julius Center for Health Sciences and Primary Care; F. L. J. Visseren, Department of Vascular Medicine; F. L. Moll, Department of Vascular Surgery; L. J. Kappelle, Department of Neurology; W. P. T. M. Mali, Department of Radiology; and P. A. Doevendans, Department of Cardiology, University Medical Center, Utrecht, The Netherlands.