Abstract
BACKGROUND: We aimed to evaluate the analytic performance of 3 rapid HIV viral load assays: the novel Xpert HIV-1 VL XC (Xpert XC), Xpert HIV-1 VL (Xpert VL), and m-PIMA HIV-1/2 VL (m-PIMA). SETTING: Two South African clinics. METHODS: We conducted a prospective diagnostic accuracy study. Site-laboratory technicians and nurses used the Xpert XC, Xpert VL, and m-PIMA to test plasma samples from people with HIV receiving antiretroviral therapy. We compared results with the Roche cobas HIV-1 reference assay. We determined accuracy to detect viraemia at the World Health Organization (WHO) failure threshold of 1000 copies/mL on all 3 assays, and 50 and 200 copies/mL on the Xpert assays. We assessed the agreement using Bland-Altman plots. RESULTS: We enrolled 140 participants (98 [70%] women, median age 37 years), who provided 189 paired samples at one or more timepoints. We tested 174 samples with the Xpert XC, 188 with the Xpert VL, and 128 with the m-PIMA. At 1000 copies/mL, sensitivity and specificity (95% confidence intervals) were 97% (82 to 100) and 98% (93 to 99) (Xpert XC), 100% (87 to 100) and 96% (91 to 98) (Xpert VL), and 92% (72 to 99) and 99% (93 to 100) (m-PIMA) respectively. At 50 copies/mL, sensitivity and specificity were 93% (81 to 98) and 96% (91 to 99) (Xpert XC), and 95% (84 to 99) and 95% (90 to 98) (Xpert VL) respectively. Mean bias was -0.10 (-0.54 to 0.34) log10 copies/mL (Xpert XC), 0.07 (-0.37 to 0.52) log10 copies/mL (Xpert VL), and -0.26 (-0.83 to 0.31) log10 copies/mL (m-PIMA). CONCLUSIONS: In these South African clinics, the accuracy of all 3 assays was clinically acceptable to detect viraemia at the WHO failure threshold, whereas both Xpert assays were also accurate at detecting low-level viraemia.