Introduction

Avascular necrosis (AVN), also known as osteonecrosis, is characterized by death of the osteocytes. Although the pathogenesis of AVN is not fully understood, the pathogenesis is believed to be multifactorial and interruption of blood supply is the main mechanism of AVN. Alcohol abuse, corticosteroid (CS) use, altered lipid metabolism, sickle cell anemia, coagulopathy, trauma, Gaucher disease, and connective tissue disorders are risk factors for the development of AVN [1]. The exact frequency of AVN is unknown. There are approximately 20,000–30,000 patients newly diagnosed with AVN each year in the United States [2]. In the German-speaking countries, AVN of hip has an incidence of 0.01% [3]. The prevalence of AVN in hip joints was reported as 0.725% in the Chinese population [4]. AVN is also observed in rheumatologic disorders, especially in systemic lupus erythematosus (SLE). The reported prevalence of AVN in patients with SLE changes between 10 and 44% [5, 6].

The clinical manifestation is variable, ranging from asymptomatic radiologic incidental finding to severe pain, joint damage and limitation in mobility and collapse of joint. Hip is the most commonly affected joint, followed by knee [7]. AVN may require surgery and need of tapering CS dose. Thus, it constitutes an additional challenge in the management of underlying disease. Furthermore, for a better outcome in AVN, early diagnosis and proper management in high risk patients is crucial. AVN by itself is underlying reason in approximately 10% of all total hip replacements [8]. CS therapy is a known major risk factor for development of AVN [9]. In a study investigating incidence of AVN in hips and knees of patients who received corticosteroid therapy for various inflammatory and non-inflammatory disorders, the incidence was 37% in systemic lupus erythematosus (SLE), 27% in polymyositis/dermatomyositis, 14% in vasculitis syndromes, 45% in adult-onset Still’s disease, 33% in systemic sclerosis, 25% in granulomatosis with polyangiitis and 0% in BD [10]. Exact incidence and prevalence of AVN in BD is largely unknown since no comprehensive study evaluated AVN in patients with BD.

BD is a chronic, multi-system, vasculitic disorder [11]. BD has acute temporary manifestations, such as oral aphthous ulcers, genital ulcers, arthritis which are usually resolve without a chronic sequela. However pan-uveitis, vascular and central nervous system (CNS) involvements usually results in severe organ damage, such as vision loss, chronic post-thrombotic syndrome, chronic progressive neuro-Behçet disease (neuro-BD) and brainstem atrophy [12]. AVN is also a chronic clinical entity observed in BD.

Although CSs are mainstay of the treatment in these major organ/system involvements of BD which has potential morbidity and mortality, AVN is rarely reported in patients with BD in the literature when compared to SLE and other rheumatologic disorders [13,14,15,16,17,18,19]. The limited published data about AVN in patients with BD may be due to masking effects of other rheumatologic symptoms like arthritis. Another reason may be due to asymptomatic nature of AVN in some patients which leads to underestimation of AVN in BD patients. To prevent underestimation of AVN in BD, it is important to define its clinical features. In this study, we evaluated clinical and treatment characteristics of nine BD patients with diagnosis of AVN.

Materials and methods

We retrospectively reviewed the medical records of 337 patients with BD who followed at the outpatient tertiary rheumatology clinic of Gazi University Hospitals between 2011 and 2018. Nine patients with diagnosis of AVN were detected. The diagnosis of BD was made according to the International Study Group (ISG) criteria [20]. Demographic characteristics, duration of disease (years), the time between development of AVN and diagnosis of BD (years), history of smoking, surgery due to AVN, the site of AVN confirmed by magnetic resonance imaging (MRI) were defined. BD patients were classified as ocular, mucocutaneous, vascular, neurologic involvement and arthritis according to their clinical manifestations. The dominant organ involvement was defined as type of involvement that primarily determined the treatment decision.

Antinuclear antibody (ANA), anti-cardiolipin (aCL) IgM and IgG (U/ml), anti-β2 glycoprotein I (aB2-GPI) IgM and IgG antibodies (U/ml), serum total cholesterol (T Chol, 0–200 mg/dl), low density lipoprotein (LDL, 0–130 mg/dl), triglyceride (TG, 0–150 mg/dl), erythrocyte sedimentation rate (ESR, 0–20 mm/h) and C-reactive protein (CRP, 0–5 mg/L) levels at the time of diagnosis of AVN were recorded. Acquired and congenital hypercoagulability tests were performed in patients with vascular involvement.

We also evaluated CS treatment of patients as following; highest daily CS dose (mg/day), age at initial CS treatment was initiated, the time interval between initial CS administration and diagnosis of AVN (months), history of pulse CS administration, cumulative CS dose (mg) before diagnosis of AVN.

Other risk factors for AVN, including trauma, alcohol consumption, hemoglobinopathy, sickle cell anemia, coagulopathy, myeloproliferative diseases, leukemia and lymphoma, HIV infection, renal failure, inflammatory bowel diseases were also evaluated.

Search strategy

A literature review was performed for manuscripts from 1978 to November 30th, 2018 to identify BD patients with AVN. The search was performed in two electronic databases (MEDLINE/PubMed and Scopus) using the following keywords: “Behçet”, “Behçet’s”, “Behcet”, “Behcet’s”, “osteonecrosis”, “aseptic necrosis”, “avascular necrosis” and “femoral head necrosis”. We included all manuscripts (case reports, randomized controlled trials, case control, cohort, or cross-sectional studies) related with BD and AVN. Exclusion criteria were: manuscripts which are not accessible in full-text, publication in languages other than English, case reports characterized by not definitely proven BD and manuscripts which do not document clinical characteristics of AVN in BD patients. After analysis, we identified seven manuscripts about AVN in Behçet’s disease. Six out of seven manuscripts were case reports and the other one was a study that evaluates the frequency of arthritis in Behçet’s disease. In latter study, three cases of AVN were reported. Aforementioned study and six case reports were included in the final analysis, in addition to our study [13,14,15, 17,18,19, 21] (Tables 1, 2).

Table 1 Summary of published cases of avascular necrosis in Behçet disease
Full size table
Table 2 Published cases of corticosteroid administration in Behçet disease patients with avascular necrosis
Full size table

Statistical analysis

Statistical analyses were performed using the SPSS software version 22. Categorical variables are presented as numbers and percentages. Continuous variables were presented as medians and ranges.

Results

All of patients in our study were male. The dominant clinical manifestation was vascular in five patients, ocular in two patients, concomitant vascular–neurologic in one and ocular–neurologic in the other patient. The median age at the onset of BD was 25 years (range 16–57) and for AVN the median age was 35 years (range 17–60). Median duration of BD was 7 years (range 4–17). Median time between diagnosis of BD and diagnosis of AVN was 3 years (range 1–16) (Table 3).

Table 3 Baseline characteristics of nine male patients with Behçet disease and avascular necrosis in our cohort
Full size table

In the present study, ANA was positive in three patients and aCL IgM and IgG and aB2-GPI IgM and IgG level was slightly above normal in two patients. Lipid profile at the time of diagnosis of AVN was available in four patients. One of these patients had elevated LDL and T. Chol levels, 204 mg/dl and 273 mg/dl, respectively. Another patient had slightly elevated TG level of 185 mg/dl. Median CRP and ESR were 13 mg/l (range 3–100) and 32 mm/h (range 3–76) at the time of diagnosis of AVN, respectively. Azathioprine (AZA) and cyclophosphamide (CYC) were the most commonly used immunosuppressive agents (Table 3). Except for one patient all were active smokers at the time of diagnosis of AVN.

We observed development of AVN in 20 joints (12 knee, 7 hip and 1 metacarpophalangeal) of nine patients. Multiple joints were involved in seven patients. Knee was the most commonly affected joint which was observed in six patients and in all of these both knees were affected from AVN. Hips were the second most commonly affected joint and the site was bilaterally affected in two patients, whereas unilateral in three of them. One patient had small joint AVN of the left third metacarpophalangeal joint along with knee and hip joints.

The median time interval between initial CS dose and AVN diagnosis was 24 months (range 2–100). The median highest daily CS dose was 64 mg/day (range 48–80) and median cumulative CS dose prior to AVN was 18 g (range 6–21). All of patients had multiple intravenous pulse steroid administration, but one patient received only a single dose of 500 mg pulse CS (Table 4). Three patients had needed of joint surgery due to AVN (Table 3).

Table 4 Summary of corticosteroid administration in nine male patients with Behçet disease and avascular necrosis in our cohort
Full size table

Discussion

In the present study, we evaluated the clinical characteristics of AVN in nine patients with BD, which has quite limited data. When compared to previous published cases, this study comprehensively evaluated clinical, treatment and laboratory characteristics and had the highest number of BD patients with AVN.

Corticosteroid use is associated with osteonecrosis in a dose-related fashion in adult patients with chronic inflammatory disorders [22]. All of patients who had had major organ involvements of BD had received CS in our cohort. There are a few case reports with use of CS prior to diagnosis of AVN in BD patients, but they were unable to provide details of the dose of CS [17, 19]. In this study, eight out of nine patients had a cumulative CS dose of more than 10 g and all of them had multiple intravenous (IV) pulse CS administrations. Most of patients also had oral daily CS dose of more than 60 mg. This suggest us dose-dependent relationship with development of AVN and CS use. Hence, pulse therapy and high dose daily CS may constitute important risk factors for development of AVN in BD patients. In one patient, the time interval between the CS treatment and AVN diagnosis was only 2 months and the administered CS dose in these 2 months was 10 g due to refractory vascular BD. The short time interval between CS administration and development of AVN can be explained with many ways. The initial high glucocorticoid dose may be associated with development of AVN, as observed in SLE [6, 23, 24]. Smoking might be another facilitating factor, because the patient had had a history of 15 pack-years smoking [25]. Another reason may be related with the type of organ involvement (vascular) of BD. In literature, one BD patient without previous use of CS had diagnosis of AVN in unilateral knee joint [18]. This may postulate different mechanisms involved in pathogenesis of AVN in BD, other than CS.

Joint involvement of AVN differs between diseases. In non-SLE diseases, AVN tends to be in femoral head, but in SLE it tends to be in knees and multifocal [5, 6]. In reported cases of BD patients, most of patients had AVN of hip joints and unifocal involvement was more frequent than multifocal involvement (Table 1). In contrast to previous reports, in our study, we observed that AVN in knee joints was higher than hip joints in patients with BD, like SLE and most of patients had multifocal involvement of joints. Thus, it may be advisable to further investigate other knee and hip joints after diagnosis of a single joint with AVN in BD, even if they are asymptomatic.

It has been suggested that smoking has contributory effect on development of AVN [25]. All patients except for one were smokers. Although there is no study that compares the frequency of AVN between smoker and nonsmoker BD patients, it may be speculated that smoking is a risk factor for AVN in BD. In our study, vascular involvement was dominant involvement in most patients with AVN. This may be due to the effect of vasculitic process on AVN or requirement of high immunosuppressive treatment including CS in these patients. All of BD patients had had immunosuppressive treatment prior to diagnosis of AVN. Immunosuppressive drugs may contribute to AVN by cytotoxic effect on osteoblasts and osteoclasts [26]. But these drugs are also important in the control of active inflammation and they have steroid sparing effects. Thus, the net effect of these drugs on development of AVN is uncertain. In previous reported case, the number of male and female patients was similar. All of patients were male in our study. This may be due to more aggressive course of BD in young male patients with requirement of more intensive treatment, including CS [27].

There is an uncertain relation of antiphospholipid antibodies with development of AVN in SLE. None of our patients had a significantly elevated antiphospholipid antibody level, similar to previous case reports [17, 18]. Despite similar articular involvement with SLE in BD patients in our cohort, the absence of female gender and aCL antibodies suggest different mechanism of action. Similarities were CS use, younger age and articular pattern.

One limitation of this study is that MRI had been performed only in symptomatic patients. When asymptomatic patients on CS treatment are evaluated for development of AVN with MRI, a higher number of BD patients with AVN might be detected.

As a conclusion, although the number of patients is relatively low and to make a clear conclusion, it may be speculated that CS treatment, smoking and vascular type of involvement may predispose to development of AVN in patients with BD. AVN should be considered in BD patients who complaint with joint paint and those receiving CS treatment. Finally, since AVN by itself has significant morbidity and influence treatment decisions it might be better to take account this complication in damage indices of BD as considered in SLE.