254   Determination of a Clinically Relevant Phenotypic Resistance "Cutoff" for Abacavir Using the PhenoSense Assay.

E. R. Lanier*¹, N. Hellmann³, J. Scott¹, M. Ait-Khaled², T. Melby¹, E. Paxinos³, H. Werhane³, C. Petropoulos³, E. Kusaba³, M. St. Clair¹, L. Smiley¹, and S. Lafon¹.
¹Glaxo Wellcome, Res. Triangle Park, NC;²Stevenage, UK; and³ViroLogic, South San Francisco, CA.

Background:Currently, the cutoffs for "sensitive" and "resistant" reported in phenotypic assays are not defined by clinical data. We sought to determine a clinically relevant phenotypic cutoff value for ABC.

Methods:Data were combined from 4 GW trials (CNA2003, 3001, 3002, 3009). Despite differences in design and patient history, all enrolled NRTI experienced patients had ABC added as a single agent to their stable background therapy. Patients in this analysis (N=120) had baseline plasma HIV-1 RNA (vRNA) >400 copies/mL (Roche Amplicor), 24 week median (range 8-24) vRNA follow-up and available baseline phenotype. Statistical methods included Fisher's exact tests of the proportions and a recursive partitioning application yielding a Bonferroni adjustedc²test for the selection of phenotype cutoffs for ABC (Musser, 1999).

Results:By Fisher's exact analysis, response to <400 c/mL or any response was significantly higher when ABC susceptibility was <4.5 fold (p< 0.001).

Recursive partitioning suggested a similar ABC cutoff: 4.4 fold for any response (p=0.001) and 4.3 fold for <400 c/mL (p=0.005).

Conclusions:These data support a clinically defined phenotypic cutoff for ABC of 4.5 fold using the PhenoSense assay, although some patients with greater reductions in susceptibility may respond.