RHEUMATOLOGY
Rheumatology 2014;53:2018–2022
doi:10.1093/rheumatology/keu233
Advance Access publication 6 June 2014
Concise report
Vascular involvement in Behçet’s syndrome: a
retrospective analysis of associations and the time
course
Koray Tascilar1, Melike Melikoglu1, Serdal Ugurlu1, Necdet Sut2, Erkan Caglar1
and Hasan Yazici1
Abstract
Objective. Some features of Behçet’s syndrome (BS) tend to go together. We aimed to explore the
association and timing of various vascular events in both the venous and the arterial vascular tree.
Methods. We conducted a chart survey on the type and time of vascular involvement of BS. The crossrelationships of involvement were assessed by phi correlation coefficients. Multiple correspondence analysis was used to identify patterns of vascular involvement. The risk of vascular recurrence was also
estimated.
CLINICAL
SCIENCE
Results. We identified 882 patients with vascular involvement among 5970 BS patients (14.7%). Deep
vein thrombosis (DVT), almost always in the legs, was the most frequent single vascular event (592/882;
67.1%). The cumulative risk of a recurrent vascular event was 38.4% at 5 years. Patients with extrapulmonary artery involvement (EPAI) were significantly older than those with venous and pulmonary artery
involvement (PAI). There were significant correlations between dural sinus thrombosis (DST) and PAI,
Budd–Chiari syndrome (BCS) and inferior vena cava syndrome (IVCS) and between IVCS and superior
vena cava syndrome (SVCS). Multiple correspondence analysis further indicated clustering of PAI, DST,
BCS, IVCS and SVCS. However, EPAI and DVT clustered separately from forms of vascular disease, the
separate clustering of the DVT being attributed to its propensity to occur solo.
Conclusion. The most common type of vascular involvement in BS is solo DVT, almost always occurring
in the legs. Various forms of venous disease in BS segregate together and PAI is included in this group.
EPAI segregates separately.
Key words: Behçet’s syndrome, arterial aneurysm, pulmonary artery aneurysm, dural sinus thrombosis, venous
thrombosis.
Introduction
There is evidence that some of the features of Behçet’s
syndrome (BS) may show associations and the presence
of such associations implicate more than one disease
mechanism in what we today call BS [1, 2]. For example,
1
Department of Rheumatology, Istanbul University Cerrahpasa
Medical Faculty, Istanbul and 2Department of Biostatistics and
Medical Informatics, Trakya University Medical Faculty, Edirne,
Turkey.
Submitted 15 July 2013; revised version accepted 4 April 2014.
Correspondence to: Melike Melikoglu, IU Cerrahpasa Tip Fakultesi, Ic
Hastalikları Romatoloji Bilim Dali, Kocamustafapasa 34098 Istanbul,
Turkey. E-mail: [email protected]
it has been shown that dural sinus involvement is associated with large vessel disease [3], while we also know
that pulmonary artery aneurysms are associated with peripheral venous thrombosis in almost 90% of patients [4].
Hence it would have been reasonable to assume that the
frequency of dural sinus thrombi (DSTs) would be
increased among patients with pulmonary artery involvement (PAI). We also had the impression that involvement
of the aorta and peripheral arteries had a different pattern
of expression than that of venous involvement [5]. To
these ends we reanalysed a large cohort of BS patients
with vascular disease that we had previously reported as
an abstract [6] to test for these two contentions and to
reassess temporal relationships in vascular disease onset
during the disease course.
! The Author 2014. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: [email protected]
Vascular involvement in Behçet’s syndrome
Methods
Data extraction
Two authors (S.U. and E.C.) reviewed the charts of all
patients registered between August 1977 and July 2006
for vascular involvement. We recorded the date of disease
onset as defined by fulfilment of the International Study
Group (ISG) criteria, the date of registration to the clinic,
the date and type of vascular events and the date of the
last follow-up visit. We also recorded the types and doses
of immunosuppressive treatments with their start and discontinuation dates. We did not include instances of superficial thrombophlebitis. Time-to-event calculations were
based on the date of disease onset and thus included
vascular events that occurred before registration to the
clinic.
Statistical analyses
Mean (S.D.) is given for variables that distributed normally;
the median and first to third quartile interval (Q1–Q3) are
given for variables with a skewed distribution. Chisquared and t-tests and analysis of variance (ANOVA) or
Kruskal–Wallis tests were used as appropriate to compare
categorical and continuous data, respectively. Temporal
relationships of vascular events were graphed and vascular recurrence was analysed using Kaplan–Meier plots.
Two-tailed P-values <0.05 were considered significant.
Phi correlation was used to explore the association between vascular event types. Since 21 unique associations
were tested in this analysis, the significance threshold was
set at 0.0024 after the Bonferroni correction.
We used multiple correspondence analysis (MCA) to
have a better overall understanding of the association between the different types of vascular events and the contribution of each vascular event type to the overall
variability. MCA is a descriptive, data reduction method
that allows simplified summaries to describe relationships
between multiple categorical variables [7]. In this analysis,
all variables are cross-tabulated and associations between variable categories are analysed based on chisquared distances. These chi-squared distances can be
visualized as a joint plot of category points, where categories that occur frequently are located near the origin,
whereas categories that generate variability are plotted
further away and the distance between points reflects
their association. In addition, discrimination measures
are calculated by squaring the component loadings.
Discrimination measures reflect the contribution of each
variable to the overall variance in each dimension.
Results
We identified 882 patients with vascular involvement out
of 5970 patient charts (14.7%) (82 females, 9.2%). Age at
disease onset was 28.1 years (S.D. 7.9) and the median
follow-up was 3.1 years (Q1–Q3: 0.5–9.1). Median time to
first vascular event was 1.4 years after fulfilling the ISG
criteria (Q1–Q3: 0–4.9 years). Most of the patients (658
patients, 74.6%) experienced their first vascular event
www.rheumatology.oxfordjournals.org
within 5 years of disease onset. However, in 91 of the
882 patients (10.5%) the first vascular event occurred
before fulfilling the ISG criteria (median 1.0 year, Q1–Q3:
0.5–2.0). The initial event in these 91 patients was mostly
deep vein thrombosis (DVT) (79/ 91, 86.8%); in addition,
there were three patients with extra-PAI (EPAI) and one
patient with PAI. No patient had DST before fulfilling the
ISG criteria.
The distribution of event numbers, age at onset of BS,
age at vascular event and time to vascular event for different types of vascular involvement are summarized in
Table 1. When we compared the time to vascular events
after disease onset, DVT and DST were earlier, with a
median time after disease onset of 1.3 and 1.2 years, respectively, whereas EPAI occurred later in the disease
course, with a median time lag of 4.9 years (P < 0.01,
Kruskal–Wallis). There was also a significant difference
with respect to age at onset of BS and age at vascular
event. Patients with EPAI were significantly older both at
disease onset [32.9 years (S.D. 7.8), P < 0.01, ANOVA] and
at the time of vascular involvement [39.4 years (S.D. 10.1),
P < 0.01, ANOVA] in comparison with other types of vascular involvement (Table 1).
Correlation of vascular events
The number and co-occurrence of different types of vascular event and results of the formal correlation analyses
are cross-tabulated in supplementary Table S1, available
at Rheumatology Online. Overall there were 592 patients
with DVT and no other type of vascular involvement, 179
patients with DVT alongside other vascular (non-DVT)
events and 111 patients who had non-DVT vascular
events only.
DST and PAI were significantly correlated, with 11 of the
35 DST patients also having PAI (31.4%, ’ = 0.14, P <
0.001). Another significant correlation was between BCS
and inferior vena cava syndrome (IVCS). Twelve of the 21
patients with Budd–Chiari syndrome (BCS) also had IVCS
(57.1%, ’ = 0.26, P < 0.001). IVCS also correlated with
superior vena cava syndrome (SVCS), with 14 of the 69
IVCS patients having SVCS (20.3%, ’ = 0.12, P < 0.001).
DVT negatively correlated with all types of vascular
involvement except with IVCS (see supplementary
Table S2, available at Rheumatology Online). In the great
majority of the patients [758/771 (98.3%)] the DVT was in
the legs (Table 1), so the designation of a DVT refers, most
of the time, to a DVT in the legs.
MCA in two dimensions shows an explained variability
of 24.4% in dimension 1 and 16.8% in dimension 2. In the
joint plot of category points we see that DVT is near the
origin, DST and PAI deviate horizontally along dimension 1
and SVCS, IVCS and BCS deviate positively in both dimensions. EPAI deviates negatively along dimension 2
and localizes at the furthest distance from the central
and peripheral venous events (Fig. 1A). The discrimination
plot shows that DVT has the largest discriminating value in
dimension 1, followed by BCS, PAI, IVCS, DST and SVCS,
whereas EPAI had minimal discriminating value. EPAI had
the greatest discriminating value on the 16% variability
2019
Koray Tascilar et al.
TABLE 1 Distribution of the type, age of onset and timing of vascular events
Type of
vascular
involvement
Peripheral veins
Arms and
legsa
DST
Central veins
SVCS
IVCS
BCS
Arteries
Pulmonary
arteriesb
Extrapulmonary arteriesc
Other
Other arterial
and venous
eventsd
Patients
(n = 882),
n (%)
Events
(n = 1358),
n (%)
771 (87.4)
961 (70.8)
35 (3.9)
Patients with
a single event,
n (%) (% within
category)
First vascular
event, n (%)
(% within
category)
Age at disease
onset, mean
(S.D.), years
Age at event,
mean (S.D.),
years
Time to
event, median
(Q1–Q3),
years
458 (59.1)
737 (95.6)
28.2 (7.9)
31.4 (8.8)
1.3 ( 0.3–4.8)
36 (2.6)
7 (20.0)
23 (65.7)
24.9 (7.5)
27.5 (7.6)
1.2 (0–4.9)
79 (8.6)
69 (7.8)
21 (2.4)
79 (5.8)
69 (5.1)
21 (1.5)
17 (21.5)
3 (4.3)
4 (19.0)
46 (58.2)
35 (50.7)
15 (71.4)
26.2 (6.8)
25.9 (9.6)
25.7 (9.0)
30.6 (7.5)
31.9 (10.0)
29.8 (10.1)
2.5 (0.3–6.3)
4.5 (0.5–10.0)
2.3 ( 0.1–8.3)
90 (10.4)
91 (6.8)
21 (23.3)
40 (44.4)
26.6 (8.3)
30.9 (9.7)
3.1 (0.4–6.0)
57 (6.4)
69 (5.1)
18 (31.6)
35 (61.4)
32.9 (7.8)
39.4 (10.1)
4.9 (0.9–9.9)
32 (3.6)
32 (2.3)
6 (18.7)
10 (31.2)
28.7 (9.5)
34.7 (10.0)
4.3 (1.0–10.2)
a
Includes 13 patients with thrombosis in the deep veins of the arms. bIncludes 14 pulmonary artery thrombosis events.
Includes 15 occlusive/thrombotic extrapulmonary arterial involvement events. dCerebral, renal, retinal, coronary/cardiac,
portal and mesenteric vascular events. SVCS: superior vena cava syndrome; IVCS: inferior vena cava syndrome; BCS:
Budd–Chiari syndrome; DST: dural sinus thrombosis.
c
explained in dimension 2, whereas DVT, PAI, SVCS and
DST had very low discriminating values in this dimension
(Fig. 1B, see supplementary Table S2, available at
Rheumatology Online).
An episode of pulmonary embolism was not described
in any of the charts. Fourteen of 90 patients with PAI had
isolated pulmonary artery thrombosis (PAT), with three
having simultaneous DVT.
Treatment and course of the vascular disease
The immunosuppressive treatment given to BS patients
with vascular involvement was AZA in 552 patients
(62.5%) with a median dose of 150 mg/day (Q1–Q3:
150–175) for a median of 1.5 years (Q1–Q3: 0.4–3.1) and
CYC in 200 patients (22.7%) with a median dose of
150 mg/day orally (Q1–Q3: 100–150) or 1 g/month i.v.
(IQR same as median) for an overall median duration of
1 year (Q1–Q3: 0.4–2.6). We also tabulated the proportion
of patients with each treatment type by decade of registration to the clinic. An increasing proportion of patients
were treated with AZA and a decreasing proportion of
patients were treated with CYC over three decades (see
supplementary Table S3, available at Rheumatology
Online).
Overall 312 of 882 patients (35.4%) had recurrent vascular events during the course. Kaplan–Meier analysis of
patients with a first vascular event showed a cumulative
new vascular event risk of 23.0% at 2 years and 38.4% at
5 years. Supplementary Fig. S1, available at
Rheumatology Online, shows the distribution of initial
and recurrent vascular event types in time. We observed
that DVT was the most common type of recurrent event
over time and that the proportion of patients with EPAI,
PAI and central vein thrombosis (IVCS, SVCS and BCS)
increased along vascular recurrences (see supplementary
Fig. S1, available at Rheumatology Online).
2020
Discussion
This chart survey first showed that isolated DVT was the
most frequent manifestation of major vascular disease in
BS. As such, it was noteworthy that it segregated as the
strongest dimension in the MCA. Our survey also showed
that among BS patients with vascular involvement, arterial
disease outside the lungs (EPAI) had a different pattern of
disease expression when compared with peripheral
venous and/or pulmonary vascular disease. An important
component of this difference was the significantly later
age of onset of EPAI, by a mean of 8.5 years, as compared with PAI (Table 1). Our results also back up our
contention that the presence of DST and PAI are significantly associated.
Up to now, major vascular involvement in BS has, in
general, been considered as either arterial—grouping pulmonary and systemic arteries together—or venous [5,
8–10]. Our survey suggests this might be unjustified.
With this survey we have found evidence that DST, PAI
and venous disease outside the lungs co-occur significantly in frequency, while such co-occurrence does not
seem to be the case with arterial lesions outside the lungs.
Furthermore, the onset of non-pulmonary arterial disease
is at a significantly later age (Table 1) than that of PAI or
DST and other types of venous involvement. It is
www.rheumatology.oxfordjournals.org
Vascular involvement in Behçet’s syndrome
FIG. 1 Multiple correspondence analysis
Joint plot of (A) category points and (B) discriminating
values of different types of vascular events in two
dimensions of variance.
important to note that this later onset for EPAI has also
been observed before by us [5] and a French group [8].
Although both are classified as elastic arteries, the aorta
and pulmonary arteries are not much alike. The pulmonary
artery is thinner, stiffer [11] and possesses a different
endothelial structure [12]. In addition, vasa vasora, the
major culprit in large vessel vasculitis, are densely distributed in the vessel wall more similar to a large vein than
what is found in the aorta [13]. Low pressure, low resistance to flow and the hypoxic blood it carries also renders
the pulmonary artery similar to veins, therefore the
www.rheumatology.oxfordjournals.org
proposed dual classification for arterial disease in BS
seems to be plausible.
There is considerable debate about what percentage of
vascular lung lesions in BS are related to pulmonary
emboli. As was the case in this survey, we have maintained that pulmonary emboli are rare in BS [4, 5, 14].
Pathologic changes, autopsy findings, differences in pulmonary perfusion scan patterns and the necessity of immunosuppression for treatment success also support this
view [8, 15–17]. While others tend to consider thrombosis
in the pulmonary vascular tree as emboli [8, 18], we saw in
a recent survey that isolated PAT is within the spectrum of
PAI [14] and would not disagree with the conclusion that it
is difficult to distinguish pulmonary emboli from that of in
situ inflammatory thrombosis of the pulmonary artery [8].
One interesting and at first glance counterintuitive result
of our survey was the negative correlation of DVT with all
other types of vascular involvement except for IVCS.
However, the negative correlations of DVT with many different forms of involvement, we reason, are due to the
large number of patients who had solitary DVT (see supplementary Table S1, available at Rheumatology Online).
This does not mean that BS patients with DVT have a
lower risk of vascular involvement elsewhere.
As was also brought up in this survey, our approach to
treatment of vascular BS is to use AZA and short-term
corticosteroids for low-risk vascular involvement such as
DVT or SVCS and CYC with high-dose corticosteroids for
high-risk vascular involvement such as pulmonary artery
aneurysms or BCS [19]. We usually do not recommend
anticoagulants for the treatment of thrombosis associated
with BS and discontinue such treatment if previously
prescribed.
Our study had some limitations. This was a retrospective chart review, thus ascertainment and sequencing of
events are subject to error. Another limitation is that we
excluded superficial thrombophlebitis. Although superficial thrombophlebitis is well known to be associated
with DVT [1], these patients frequently present with nodular lesions that are clinically indistinguishable from erythema nodosum [20] and the presence of an underlying
thrombophlebitis is not investigated in every case.
In conclusion, our survey provides additional evidence
that venous pathology in BS is a spectrum that includes
dural sinuses and pulmonary arteries. On the other hand,
extrapulmonary major arterial involvement seems not to
be included in the said disease spectrum.
Rheumatology key messages
Isolated deep vein thrombosis is the most common
form of major vascular disease (67.1%) in Behçet’s
syndrome.
. Various forms of venous disease cluster in Behçet’s
syndrome; pulmonary artery involvement clusters
with this group.
. Extra- pulmonary arterial involvement in BS segregates separately from venous disease and pulmonary arterial involvement.
.
2021
Koray Tascilar et al.
Disclosure statement: The authors have declared no
conflicts of interest.
9 Tursen U, Gurler A, Boyvat A. Evaluation of clinical findings according to sex in 2313 Turkish patients with
Behcet’s disease. Int J Dermatol 2003;42:346–51.
Supplementary data
10 Saba D, Saricaoglu H, Bayram AS et al. Arterial lesions in
Behcet’s disease. Vasa 2003;32:75–81.
Supplementary data are available at Rheumatology
Online.
11 Azadani AN, Chitsaz S, Matthews PB et al. Biomechanical
comparison of human pulmonary and aortic roots. Eur J
Cardiothorac Surg 2012;41:1111–6.
References
1 Tunc R, Keyman E, Melikoglu M, Fresko I, Yazici H. Target
organ associations in Turkish patients with Behcet’s
disease: a cross sectional study by exploratory factor
analysis. J Rheumatol 2002;29:2393–6.
12 Aird WC. Phenotypic heterogeneity of the endothelium: I.
Structure, function, and mechanisms. Circ Res 2007;100:
158–73.
13 Cliff WJ. Blood Vessels. New York, NY, USA: Cambridge
University Press, 1976.
2 Yazici H, Ugurlu S, Seyahi E. Behcet syndrome: is it one
condition? Clin Rev Allergy Immunol 2012;43:275–80.
14 Seyahi E, Melikoglu M, Akman C et al. Pulmonary artery
involvement and associated lung disease in Behcet disease: a series of 47 patients. Medicine 2012;91:35–48.
3 Tunc R, Saip S, Siva A, Yazici H. Cerebral venous
thrombosis is associated with major vessel disease in
Behcet’s syndrome. Ann Rheum Dis 2004;63:1693–4.
15 Lakhanpal S, Tani K, Lie JT et al. Pathologic features of
Behcet’s syndrome: a review of Japanese autopsy registry
data. Hum Pathol 1985;16:790–5.
4 Hamuryudan V, Yurdakul S, Moral F et al. Pulmonary
arterial aneurysms in Behcet’s syndrome: a report of
24 cases. Br J Rheumatol 1994;33:48–51.
16 Matsumoto T, Uekusa T, Fukuda Y. Vasculo-Behcet’s
disease: a pathologic study of eight cases. Hum Pathol
1991;22:45–51.
5 Kural-Seyahi E, Fresko I, Seyahi N et al. The long-term
mortality and morbidity of Behcet syndrome: a 2-decade
outcome survey of 387 patients followed at a dedicated
center. Medicine 2003;82:60–76.
17 Palla A, Pazzagli M, Manganelli D et al. Resolution of
pulmonary embolism: effect of therapy and putative age
of emboli. Respiration 1997;64:50–3.
6 Melikoglu M, Ugurlu S, Tascilar K et al. Large vessel involvement in Behçet’s syndrome: a retrospective survey.
Ann Rheum Dis 2008;67(Suppl II):67.
7 Sourial N, Wolfson C, Zhu B et al. Correspondence analysis is a useful tool to uncover the relationships among
categorical variables. J Clin Epidemiol 2010;63:638–46.
8 Desbois AC, Wechsler B, Resche-Rigon M et al.
Immunosuppressants reduce venous thrombosis relapse
in Behcet’s disease. Arthritis Rheum 2012;64:2753–60.
2022
18 Saadoun D, Wechsler B, Resche-Rigon M et al. Cerebral
venous thrombosis in Behcet’s disease. Arthritis Rheum
2009;61:518–26.
19 Hatemi G, Silman A, Bang D et al. EULAR recommendations for the management of Behcet disease. Ann Rheum
Dis 2008;67:1656–62.
20 Misago N, Tada Y, Koarada S, Narisawa Y. Erythema
nodosum-like lesions in Behcet’s disease: a clinicopathological study of 26 cases. Acta Derm Venereol 2012;
92:681–6.
www.rheumatology.oxfordjournals.org