Objectives To study the standardized incidence ratio (SIR), time trend and risk factors of AVN in patients with SLE.
Methods The records of all patients who fulfilled ≥4 ACR criteria for SLE in our unit between 1999 and 2014 were reviewed. Patients who developed AVN at any sites ever since the diagnosis of SLE were identified. A group of SLE controls who did not have evidence of AVN were randomly selected from our cohort database in a 4:1 (control/case) ratio, matched for age, sex and SLE duration. The SIR of AVN in SLE and its time trend was calculated by data retrieved from our hospital clinical information registry and the Census data from our Government. Risk factors for AVN in SLE were studied by multivariate logistic regression. The following factors were considered to be covariates in the regression model: hypertension, diabetes mellitus, lipid level, previous septic arthritis, maximum daily dose and cumulative dose of prednisolone ever used, cushingoid body habitus, cutaneous vasculitis, Raynaud's phenomenon, antiphospholipid antibodies, and a propensity score derived from a separate logistic regression model for the probability of use of high-dose prednisolone (>0.8mg/kg/day) for different SLE manifestations.
Results 55 patients with symptomatic AVN (87%women;age 33.4±12.4 years; SLE duration 61.2±62.2 months) and 220 matched SLE controls (87% women, age 34.3±10.6 years; SLE duration 71.7±50.1 months) were studied. The point prevalence of AVN in our SLE cohort (N=743) was 7.4%. All the patients with AVN had been treated with glucocorticoids (GCs). Compared to controls, AVN patients had used a significantly higher cumulative doses of prednisolone (16.5±14.6 vs 10.7±11.3 grams; p=0.003). A total of 104 sites of AVN were diagnosed in 55 patients (69%≥2 sites). The hip was most commonly affected (82%), followed by the femoral condyle (9%) and the humeral head (5%). Bilateral involvement was present in 67% of the patients. Surgical treatment (core decompression, vascularized bone graft or joint replacement) was performed for 41% of the AVN lesions. The age and sex stratified SIRs of AVN in our SLE patients was 131 (86.6-199; p<0.001) in the period 1995-2004 and 56.0 (34.3-91.4; p<0.001) in the period 2005-2014. In both decades, the age stratified SIR was highest in the youngest age group (<19 years of age) with figures of 1161 (357-3770; p<0.001) in 1995-2004 and 778 (267-2270; p<0.001) in 2005-2014, respectively. Logistic regression analysis revealed the following factors independently associated with the occurrence of AVN, adjusted by the propensity score for high-dose prednisolone: preceding septic arthritis of the involved joint (odds ratio [OR] 15.4[1.3-181.2]; p=0.03), Cushingoid body habitus (OR 2.3[1.0-5.1]; p=0.043), LDL-cholesterol (OR 1.4[1.0-2.0]; p=0.041), maximum daily dose of prednisolone (mg/kg) (OR 6.0[1.2-30.7]; p=0.031) and cumulative dose of prednisolone in the first 6 months of treatment of a SLE flare (OR 1.4[1.0-1.8]; p=0.047).
Conclusions AVN is prevalent in SLE patients, particularly in younger patients. GC use, Cushingoid body habitus, serum LDL-cholesterol level and previous septic arthritis are independently associated with AVN. There is a trend of reduction in the SIR of AVN in our SLE patients over the past 2 decades, which is probably attributed by the judicious use of GCs and the early administration of GC-sparing agents.
Disclosure of Interest None declared