Abstract
INTRODUCTION: Sero-epidemiological studies of herpes simplex virus (HSV) type 2 infection in Africa remain difficult to interpret as a result of the high rate of false-positive results observed when using the new recombinant gG2 HSV-2 ELISA tests. The performance of two widely used gG2 ELISA was compared to derive an appropriate testing algorithm for use in South Africa. METHODS: Sera from 210 women attending family planning clinics in Johannesburg were tested using HerpeSelect and Kalon HSV-2 gG2 assays. Sera from 20 discordant pairs, 44 concordant positive and 33 concordant negative samples were further tested by HSV Western blot. The sensitivity and specificity of each test and of combination algorithms compared with Western blot were calculated. RESULTS: HerpeSelect had a sensitivity of 98% (95% CI 95 to 100) and specificity of 61% (95% CI 48 to 74). Kalon was less sensitive (89%, 95% CI 83 to 94) but more specific (85%, 95% CI 61 to 100). Seroprevalence may have been overestimated by as much as 14% by HerpeSelect. Specificity was improved by raising the cut-off index for the determination of a positive result for HerpeSelect (to >or=3.5), but not for Kalon. HIV-1 infection reduced the specificity of HerpeSelect to 30%. Improved sensitivity and specificity were obtained by a two-test algorithm using HerpeSelect (>or=3.5) as the first test and Kalon to resolve equivocal results (sensitivity 92%, 95% CI 82 to 98; specificity 91%, 95% CI 79 to 98). CONCLUSION: Newer HSV-2 serological tests have low specificity in this South African population with a high HIV-1 prevalence. Two-step testing strategies could provide rational testing alternatives to Western blot.