Voxel-based morphometry of brain SPECT can detect the presence of active central nervous system involvement in systemic lupus erythematosus.

OBJECTIVE: To determine the value of voxel-based morphometry (VBM) of brain SPECT (single-photon emission computed tomography) images (BSI) in discriminating active central nervous system (CNS) manifestations in systemic lupus erythematosus (SLE) patients.

PATIENTS AND METHODS: Forty SLE patients (mean age 33 yrs) and 33 normal volunteers were submitted to BSI. SLE patients were screened for the presence of CNS involvement following the American College of Rheumatology (ACR) case definition. Patients with CNS infections, uraemia, diabetes and previous ischaemic or haemorrhagic stroke were excluded. Magnetic resonance imaging (MRI) scans were obtained in a 2T scanner (Elscint Prestige) with T1- and T2-weighted images. BSI were performed after injection of 1110 MBq (30 mCi) of (99m)Tc-ECD (ethyl-cysteinate-dimer). BSI were analysed using the statistical parametric mapping. After normalization, segmentation and smoothing the groups of SLE patients with active and inactive CNS manifestations and healthy volunteers were compared using VBM. Post-processed images were compared voxel-by-voxel using t-test in order to determine differences of intensity between groups. This analysis included grand mean scaling, proportional threshold masking (set to 0.4) and implicit masking. A P-value of 0.001 and cluster size of 32 were taken into consideration.

RESULTS: VBM analyses of BSI did not show any differences between SLE patients with inactive CNS involvement and normal controls. However, the group of SLE patients with active CNS involvement had a global hypoperfusion, more intense in the frontal, dorsolateral and medial temporal lobe when compared with SLE patients without CNS involvement (P = 0.001) and healthy volunteers (P = 0.001).

CONCLUSION: VBM of BSI is a useful and objective method for detecting perfusion abnormalities in SLE patients, which is indicative of active CNS involvement. However, it is not helpful in differentiating the clinical sub-types of CNS involvement according to the ACR classification.