Antivirogram or phenosense: a comparison of their reproducibility and an analysis of their correlation

The Antivirogram and PhenoSense assays are widely used phenotypic tests for HIV drug resistance. There are limited data on the reproducibility of each assay, and little is known about the correlation between the two. Using data from the Stanford HIV drug resistance database, we performed a comprehensive analysis of the reproducibility of each assay, and calculated the correlation and concordance of the two assays using both general IC50 fold change cutoff values and drug-specific cutoff values. Although the within-assay correlations were high (rank correlation coefficients r=0.94 and r=0.95 for the Antivirogram and PhenoSense assays, respectively), the between-assay correlation was considerably lower (r=0.36). Using drug-specific cutoff values for viruses classified as resistant by the Antivirogram or PhenoSense assays, respectively, only 71.4% [95% confidence intervals (95% CI): 58.7-82.1%] and 57.0% (95% CI: 45.3-68.1%) of the samples were classified as resistant using the other assay. The poor agreement between the assays was primarily due to the extremely poor correlation between these assays for samples with low resistance values (r=0.02 and r=0.61 for samples with the Antivirogram measurements lower or higher than 2.0, respectively). Since the cutoff values for both assays are relatively low, our analysis suggests that one should be very careful when interpreting measurements that are near the cutoff values for drug resistance.