Coronary atherosclerosis and cardiovascular mortality in hemophilia


Margaret V. Ragni, Division of Hematology/Oncology, University of Pittsburgh Medical Center, Hemophilia Center of Western Pennsylvania, 3636 Boulevard of the Allies, Pittsburgh, PA 15213-4306, USA.
Tel.: +1 412 209 7288; fax: +1 412 209 7281.

Although mortality in individuals with hemophilia is 2- to 3-fold higher than the general population, ischemic heart disease mortality is 50–80% lower than the general population [1–5]. Moreover, hemophilia carriers whose factor (F)VIII levels are low, but not deficient, also appear to have low death rates of ischemic cardiac disease [6]. Elevated circulating FVIII levels [7], as well as fibrinogen [8] and von Willebrand factor [9], are considered risk factors for cardiovascular disease. Yet, it is possible that low FVIII levels may reduce the likelihood of clot formation, even in the presence of cardiovascular risk factors, such as smoking or elevated cholesterol, and, thereby, decrease coronary thrombosis and myocardial infarction in individuals with hemophilia.

In atherosclerotic-prone apolipoprotein E-FVIII double knock-out mice, it is of interest, that despite higher cholesterol levels, atherosclerotic aortic plaques are similar to those in apolipoprotein E mice with normal FVIII levels [10]. In humans, it remains controversial whether the severity of FVIII deficiency, < 0.01 U mL−1 (severe) vs. ≥ 0.01 U mL−1 (moderate, mild) correlates with ultrasound intima-media thickness. Some affirm that hereditary bleeding disorders such as hemophilia and von Willebrand disease are protective [11–13], whereas others disagree [14,15], but, to date, no study has evaluated the extent of coronary vessel stenosis histopathologically.

We, therefore, conducted a cross-sectional study to compare intraluminal coronary stenosis at autopsy and risk factors for ischemic heart disease in 14 hemophilic cases and 42 age-, gender- and race-matched non-hemophilic controls. Cases were identified as all men with hemophilia A or B previously cared for at the Hemophilia Center of Western Pennsylvania (HCWP) on whom autopsies were available. The dates of death in cases were between 1983 and 1992. Controls were identified as the first three consecutive age-matched (within 5 years), gender-matched, race-matched men without hemophilia, on whom autopsy results were available, beginning in 2007, through the University of Pittsburgh Medical Center (UPMC) anatomic pathology computerized electronic medical record. The date of death in controls was between 1994 and 2007. Autopsies were linked to medical records in the UPMC Medical Archival Records System (MARS) through an independent mediator or ‘honest broker’ who de-identified the medical records to maintain confidentiality of research subjects’ private information. Autopsy record review was approved as an exempt study by the University of Pittsburgh Institutional Review Board.

Autopsy material from the cases and controls, including all hematoxylin and eosin (H&E) slides taken from the coronary vessels identified through the UPMC anatomic pathology computerized system, was reviewed by one of the authors (L.N.). Intraluminal coronary stenosis was evaluated on autopsy slides, using a semi-quantitative scoring system with 0 = minimal (< 25%), 1 = mild (25–49%), 2 = moderate (50–74%) and 3 = severe (≥ 75%) stenosis.

Clinical data were obtained from de-identified medical records on cases, from the Hemophilia Center of Western PA, and on controls, from the UPMC Medical Archival Records System (MARS), to identify cause of death and assess risk factors for ischemic heart disease. The risk variables assessed included blood pressure, cholesterol, elevated creatinine, history of diabetes, history of smoking and body mass index (BMI). Hypertension was defined as a systolic blood pressure > 140, and/or diastolic blood pressure > 90, documented history of hypertension, or need for anti-hypertensive medication. Hypercholesterolemia was defined as total cholesterol > 200 mg dL−1, documented history of hypercholesterolemia, or need for cholesterol-lowering medication. Diabetes was defined as known history of diabetes. Smoking was defined as any history of smoking ever. Coronary symptoms were defined as chest pain, angina or myocardial infarction. Renal insufficiency was defined as a creatinine > 1.2 gm dL−1. Overweight was defined as a BMI ≥ 25, and obesity was defined as BMI ≥ 30. Prophylaxis was defined as 2–3 times weekly FVIII infusion to prevent hemorrhages. Data for each variable were established by the variable taken in time closest to death. No correction was made for missing data, which were noted by corrected denominator for each variable reported.

Data were analyzed by descriptive statistics, to identify the mean, median, range, minimum and maximum values for each variable. Study size was determined by the availability of autopsies in cases, n = 14, and to maximize statistical power, three controls were chosen for each case, matched by age within 5 years, race and gender. The degree of coronary luminal stenosis was compared with cardiovascular risk factor frequency between cases and controls.

The mean coronary artery stenosis score did not differ between hemophilic cases and non-hemophilic controls (Table 1). Similarly, the proportion with cardiovascular risk factors, including hypertension, hypercholesterolemia, smoking and diabetes did not differ between cases and controls. The frequency of coronary symptoms and mean age at death were similar between groups, although fewer hemophilic men succumbed to cardiopulmonary death. These results were unchanged when controls with cardiovascular symptoms were omitted (Table 1). All cases had hemophilia A, 10 with severe disease (FVIII ≤ 0.01 U mL−1), two with moderate disease (FVIII=0.02, 0.04 U mL−1) and two with mild disease (FVIII=0.07, 0.08 U mL−1). Although HIV infection was more common in those with severe disease, coronary stenosis, cardiovascular risk factors and coronary symptoms were not related to hemophilia severity.

Table 1.   Cardiovascular risk factors and coronary stenosis scores in hemophilic cases and non-hemophilic controls
 Hemophilic cases
(n = 14)
Non-hemophilic controls
(n = 42)
Non-hemophilic controls without CVD
(n = 38)
  1. Coronary stenosis score was determined at autopsy (see text). CVD, cardiovascular disease; BMI, body mass index.

Clinical findings
 Age at death (range, years)40 ± 16 (19.74)41 ± 15 (18.75)38 ± 13 (18.74)
 History HIV infection10/14 (71.4%)0/42 (0%)0/38 (0%)
 History coronary disease0/14 (0%)2/42 (4.8%)0/38 (0%)
Cause of death
 AIDS, HCV, CNS bleed13/14 (92.8%)0/42 (0%)0/38 (0%)
 Cancer, cardiac, infection1/14 (7.1%)42/42 (100%)38/38 (100%)
Organ failure, other
 AIDS7/14 (50.0%)0/42 (0%)0/38 (0%)
 Hepatitis C4/14 (28.6%)0/42 (0%)0/38 (0%)
 CNS bleed2/14 (14.3%)0/42 (0%)0/38 (0%)
 Cancer1/14 (7.1%)4/42 (9.0%)4/38 (10.5%)
 Cardiac0/14 (0%)14/42 (33.3%)11/38 (29.0%)
 Infection0/14 (0%)13/42 (30.9%)12/38 (31.6%)
 Organ failure0/14 (0%)4/42 (9.0%)4/38 (10.5%)
 Other0/14 (0%)7/42 (16.7%)7/38 (18.4%)
Cardiovascular risk factors
 Hypercholesterolemia5/14 (35.7%)8/38 (21.1%)7/34 (20.6%)
 Smoking5/14 (35.7%)15/37 (40.5%)13/33 (39.4%)
 Hypertension4/14 (28.6%)13/38 (34.2%)10/34 (29.4%)
 Overweight (BMI ≥ 25)3/14 (21.4%)28/42 (66.7%)25/38 (65.8%)
 Elevated creatinine1/14 (7.1%)13/39 (33.3%)10/35 (28.6%)
 Coronary symptoms0/14 (0%)4/42 (9.5%)0/38 (0%)
 Diabetes0/14 (0%)9/38 (23.7%)8/34 (23.5%)
Coronary stenosis score
 0 = Minimal (< 25%)3/14 (21.4%)17/42 (40.5%)17/38 (44.7%)
 1 = Mild (25–49%)9/14 (64.3%)7/42 (16.7%)7/38 (18.4%)
 2 = Moderate (50–74%)0/14 (0%)9/42 (21.4%)9/38 (23.7%)
 3 = Severe (≥ 75%)2/14 (14.3%)9/42 (21.4%)5/38 (13.2%)
 Any stenosis (≥ 25%)11/14 (78.6%)25/42 (59.5%)21/38 (55.3%)
 Stenosis score (mean ±SD)1.1 ± 0.91.2 ± 1.21.1 ± 1.1

In contrast to past studies utilizing B-mode ultrasonography which may be limited by differences in shear stress and hemodynamics in different vascular beds [1,2,6,16] and the lack of sensitivity for early intima-media thickness [17,18], our study is the first to directly assess the extent of pathologic atherosclerosis in coronary vessels in hemophilic men and non-hemophilic controls. Further, in contrast to past studies of cardiovascular risk in hemophilic men, we compared hemophilic cases with age-, race-, and gender-matched non-hemophilic controls without age restriction. In agreement with some [14], but not other studies [12,16] of cardiovascular disease (CVD) in individuals with bleeding disorders, we found that intraluminal stenosis was similar between these groups.

How FVIII deficiency protects against ischemic heart disease mortality, however, remains controversial. Thrombus formation on a ruptured atherosclerotic plaque and subsequent coronary vessel occlusion is the most common pathogenesis of transmural acute myocardial infarction and most likely to be fatal [19]. The importance of this sequence of events involving activation of both the extrinsic [20] and intrinsic [21] pathways of coagulation, is underscored by the association of elevated FVIII levels with increased risk for ischemic coronary disease [9], and reduced FVIII levels, as in congenital hemophilia A, with lower risk [3,5,22]. The observation that hemophilia carriers are protected from ischemic heart disease [6], further suggests that even minimal reduction in FVIII may protect from ischemic heart disease. Coronary atherosclerosis, however, may be fatal not only from thrombosis, but also from plaque rupture sending off atheroemboli, from coronary vasospasm that can rapidly narrow a lumen, or from intraplaque hemorrhage that can rapidly enlarge a lesion. If intraplaque hemorrhage were commonly fatal, individuals with hemophilia might be expected to have greater, not lower, fatality from coronary atherosclerosis. As most of the hemophilia patients (11 of 14) in this series died of HIV or hepatitis before they could develop heart disease, it is not possible to determine whether FVIII deficiency reduces ischemic disease mortality.

There are several limitations to this study. First, there are temporal differences between the year of death in cases and in controls, which could introduce bias. Further, the co-morbidity of the cases, and the likely unhealthy, unrepresentative control population, 60% of whom died as a result of infection or cardiac disease at the mean age of 40 years, makes comparison difficult. Second, the majority of hemophilia cases but no controls had HIV infection. Although highly active antiretroviral therapy (HAART), which may increase cholesterol and cardiovascular disease risk could potentially confound the findings of this study, this is unlikely as the HIV-infected cases had died by the time HAART therapy was introduced in 1996. A third potential limitation is the small sample size, related to the past small number of autopsies. Fourth, because hemophilia cases receive close, comprehensive care at hemophilia treatment centers, they might have had closer, more frequent medical care and/or earlier recognition or treatment of cardiovascular risks than controls. Despite this, risk factors did not differ between groups, suggesting differences in medical management did not affect study findings. Fifth, pre-existing cardiovascular disease could bias results in controls. However, the analysis was unchanged after leaving out controls with cardiovascular symptoms. Finally, data from medical files are subject to the bias associated with such studies, including missing data, records and/or differences in chronology.

Despite these limitations, our study provided an opportunity to compare pathologic coronary vessel occlusion and cardiovascular risk factors in hemophilic men with non-hemophilic controls, and demonstrates no significant differences between these groups. Larger studies will be needed to confirm these findings, now that hemophilic men are living a normal lifespan and may be exposed to longer duration, higher level clotting factor, e.g. prophylaxis.


M.V. Ragni and C.J. Foley designed the research, performed the research, and collected the data. L. Nichols collected and analyzed the autopsy and pathologic material. M.V. Ragni, K. Jeong, and C.G. Moore performed statistical analysis. M.V. Ragni, C.J. Foley and L. Nichols analyzed the data, formulated the conclusions and wrote the paper.


The study was supported by HHS Federal Region III Hemophilia Treatment Centers, Grant #1-H-30-MC-00038-01, Center for Diseases Control Prevention of Complications of Hemophilia Grant U10DD000193, the Pennsylvania Department of Health State SAP #04100000330 (M.V. Ragni), and CTRC/CTSI grant: NIH/NCRR/CTSA UL-1 RR024153.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.