Background: Amplification based HIV-1 viral load and resistance assays are expensive, technologically complex and may be difficult to implement in resource limited settings. Cavidi Exavir viral load and phenotype assays based on quantification of reverse transcriptase (RT) activity may be alternatives.

Methods: Evaluation of agreement between Cavidi RT (Y₁) and Roche HIV-1 RNA PCR (Y₂) assays (log₁₀ RNA c/mL) was based on a sample of 155 subjects enrolled in ACTG #201 (n=22), #307 (n=85) or the UNC CFAR HIV Cohort (UCHC)(n=48). The analysis relied on bivariate linear models fitted via a supplemental-expectation-maximization (S/EM) algorithm to cope with left/right censoring. Effects of mutations in RT on Roche and Cavidi assays (UCHC n=48) were explored in two ways: using mutation indicators as covariates in bivariate linear models; using logistic regression to assess association between mutations and the probability of |Y₁-Y₂|<¼ log. Cavidi phenotype assay was contrasted to efavirenz (EFV) associated mutations (UCHC n=26).

Results: For 80% of 155 cases the two assays differed by <1 log and for 47% by <¼ log. The standard deviations for Cavidi and Roche were 1.34 and 1.13 logs, and the means were 3.97 and 4.15, respectively, with a difference of 0.18 logs (SE)=0.04. For any given Roche assay value, the predicted mean deviation was E[Y₁-Y₂]=0.08(Y₂‑6.37). Correlation between Y₁ and Y₂ was r=0.91 (SE= 0.15). Of 15 samples that were <400, and 22 <1000 c/mL, with Roche, 93% and 100% were <400, and <1000, by Cavidi, respectively. Of 23 samples <400, and 40 <1000 c/mL, with Cavidi, 61% and 55% were <400, and <1000, by Roche, respectively. RT mutations did not influence the relationship between Y1 and Y2. EFV phenotypes by Cavidi were consistent with RT genotypes; no EFV mutations in patients without EFV resistance (n=9) and a mutation in each patient with resistance (n=17)(K103N n=10, G190A/S n=3, Y188L n=2, Y181C n=2). Of 7 patients without EFV use, 2 were resistant to EFV (both with K103N). Of 19 patients with prior EFV use, 15 were EFV resistant (all with EFV mutations) and 4 were EFV sensitive (all without EFV mutations).

Conclusions: Cavidi RT results were similar to Roche Amplicor for plasma HIV-1 RNA quantification. EFV phenotyping results were encouraging and may be very useful in developing world settings where initial NNRTI-based regimens are standard of care. Cavidi is also less expensive and technologically simpler than the Roche assay.