Human T cell leukemia virus (HTLV-I/II) serodiagnostic testing: disparate results among a cohort of intravenous drug users.

Three hundred and forty-six sera collected over a 2-year period from 154 San Francisco IV drug users were subjected to HTLV-I/II RIPA, Western blot (WB), Du Pont ELISA, and p24 radioimmunoassay (RIA). Tests were performed at separate sites and code not broken until study end. RIPA-positive and -negative controls consisted of Japanese adult T cell leukemia patients, healthy blood donors, and non-IV drug-using HIV-positive men. RIPA identified HTLV-I/II-positive sera not identified by the other tests. Positive RIPAs were noted in 43% of negative ELISAs (n = 279), 34% of negative WBs (n = 243), and 40% of negative RIAs (n = 270). Seventy-two sera were negative by all 3 assays, but were RIPA positive. All sera positive by RIA (n = 76) and WB (n = 67), and 66 of 67 by ELISA, were positive by RIPA. Thirty-five of 36 indeterminate WBs were RIPA positive. Seven samples indeterminate by RIPA were negative by WB and RIA; one of seven was positive by ELISA. In all instances, samples negative by RIPA (n = 154) were ELISA, p24 RIA, and WB negative or indeterminate. We conclude that when studying HTLV-I/II-endemic cohorts, screening ELISA or RIA followed by confirmatory WB or RIPA only of seropositive samples may result in a substantial number of undetected cases. Additional studies focusing on performance characteristics of serodiagnostic tests are needed to ensure accurate inferences are made in assessing HTLV-I/II prevalence rates among high-risk groups. The RIPA may be a uniquely sensitive assay to detect HTLV-I/II gene-encoded products.