Abstract
The anti-human T-cell leukemia virus type 1 (HTLV-1) antibody assay in common use has changed from the particle agglutination (PA) method to chemiluminescent immunoassay (CLIA) and chemiluminescent enzyme immunoassay (CLEIA). These assays were validated in serum but not in cerebrospinal fluid (CSF). However, anti-HTLV-1 antibody positivity in CSF is a requisite for diagnosing HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP). We qualitatively compared the assays in CSF from 47 HAM/TSP patients diagnosed using PA, 15 HTLV-1 carriers (HCs), and 18 negative controls. In determining the positivity or negativity of CSF anti-HTLV-1 antibodies, we used serum cutoff points for CLIA and CLEIA because CSF cutoff points had not been decided. Truth table analysis revealed that the performance of CLIA was closer to that of PA and that CLEIA had low sensitivity. CSF antibodies from HAM/TSP patients were all positive by PA and CLIA but 83.0% positive by CLEIA. CSF antibodies from HCs were positive in 73.3%, 80.0%, and 6.7% by PA, CLIA, and CLEIA, respectively. Receiver operator characteristic curve analysis for CSF revealed that with the default cutoff point used for serum, CLIA and PA had comparable performances and CLEIA was less sensitive. The best performances of CLIA and CLEIA with adjusted cutoff points were 94.8% sensitivity and 95.5% specificity and 89.7% sensitivity and 95.5% specificity, respectively. We conclude that low-sensitivity CLEIA can underdiagnose HAM/TSP and that CLIA is a better alternative to PA in anti-HTLV-1 antibody assay for CSF with the current cutoff points.