Background:  The Cavidi Exavir viral load and drug-resistance phenotype assays are based on measuring HIV-1 reverse transcriptase (RT) activity and have been proposed for use in resource limited settings since they are less expensive and easier to use than currently available assays.

Methods:  We measured plasma HIV RNA levels using Roche HIV-1 Monitor ver.1.5 and HIV RT activity using Cavidi Exavir Load ver.2 in 60 patients from the UNC CFAR HIV Cohort. Genotypic sequencing was performed using HIV Genosure, and mutations associated with reduced nevirapine (NVP) susceptibility were based on the International AIDS Society (IAS) Guidelines. We compared the Cavidi PhenoSense results to Genosure genotyping, and the Cavidi viral load to Roche viral load assay results.

Results:  At the time of sample collection 11 patients were taking NVP, 29 had previously received NVP, and 20 had never received NVP (with 9%, 41%, and 25% having current/prior efavirenz (EFV) exposure, respectively). Cavidi phenotype results were unavailable for 15 samples (25%) due to low viral load (n = 13) and deviations from linearity (n = 2). Among these 15 samples the median Cavidi viral load was 1168 copies/mL (IQR <400, 1572). In 45 patients with Cavidi phenotype results, 4 patients were NVP susceptible and without NVP mutations. Of the 41 patients phenotypically resistant to NVP, 78% (n = 32) also had evidence of at least 1 NVP mutation, most commonly K103N (n = 20), Y181C (n = 19), G190A (n = 8), and V108I (n = 5). The probability of a false negative result for the Cavidi phenotype compared to genotype result was 0% (n = 0/32); however, the probability of a false positive result was 69% (n = 9 of 13). Of the 9 patients with Cavidi NVP resistance and no NVP mutations detected, only 1 had previously received NVP, and none had been exposed to EFV. The viral load means for Cavidi and Roche were estimated to be 4.24 and 4.52 log copies/mL, and the SD were 1.03 and 0.83, respectively, with a difference of 0.29 log copies/mL, with standard error (SE) = 0.08. In 83% of cases the assays differed by <1 log, and for 45% by <0.25 log. The linear correlation between the 2 assays was r = 0.97 (SE = 0.10; p <0.0001). We did not see an association between NVP exposure or RT mutations and the relationship between Cavidi HIV RT and Roche HIV RNA results.

Conclusions:  The Cavidi viral load assay performed well compared to the Roche RNA assay. The Cavidi NVP phenotype results were very sensitive but had poor specificity. The Cavidi viral load assay may be a good alternative for use in resource limited settings.