HIV Drug Resistance Tests: An Update on Methods for Calculating Phenotypic Fold Change from a Viral Genotype

Tothe Editor— In a recently published article, Hirsch et al. [1] reviewed HIV drug resistance assays used in clinical practice and described a “virtual phenotype” that correlates genotypic data on the plasma HIV-1 RNA of a candidate gene with a large database of paired phenotypes and genotypes. Although the authors cite other bioinformatics systems that generate a calculated fold change (FC), our comments are specific to the Virco Type HIV-1. Hirsch et al. [1] state that virtual phenotype resistance interpretations are limited by a methodology that relies on matches that are based on preselected codons and not on the entire nucleotide sequence and that the “predictive power depends on the number of matched datasets available” [1, p. 274], with high variation for newer drugs with smaller datasets. We would like to note that the matching system used to calculate the FC in their article is no longer the methodology used by the Virco assay. Since July 2006, Virco's bioinformatics engine has been used to calculate the FC for a given sample by the following method [2]. First, linear regression modeling is performed periodically to analyze the relationship between genotype and phenotype in the Virco correlative database, which to date, has >53,000 samples with paired genotypic and phenotypic data. Significant mutations and mutation pairs that affect phenotypic susceptibility to each drug are identified, and their negative or positive impact on the FC is quantified by a resistance weight factor. Second, all of the mutations in a sample genotype are compared on a drug-by-drug basis to the current list of resistance weight factors for the drug. An FC score is then generated by calculating the sum of the values for all resistance weight factors identified in the sample genotype. This methodology is unlike the first generation of the virtual phenotype, which sought to identify matches to viruses with very similar mutational profiles. With the current linear modeling engine, accurate FC values can be reported regardless of whether there are viruses with similar mutational profiles in the database. Potential conflicts of interest. N.A.B. is the Director of Medical Affairs USA/Canada for Virco Lab. L.B. is the Vice President of Clinical Virology for Virco Lab. J.V. is the Senior Director of Global Medical Affairs for Virco BVBA.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.