icon-    folder.gif   Conference Reports for NATAP  
 
  5th European HIV Drug Resistance Workshop (EHDRW 2007),
Cascais, Portugal, March 28-30, 2007
Back grey_arrow_rt.gif
 
 
 
Darunavir Phenotype Predicts Response Better Than Genotype
 
 
  5th European HIV Drug Resistance Workshop
Cascais, Portugal
March 30, 2007
 
Mark Mascolini
 
Virco researchers determined that their vircoTYPE continuous phenotypic susceptibility score predicted responses to darunavir in the POWER trials better than three genotype-based tools [1]. Among the three genotype scores, the one proposed by Tibotec, darunavir's maker, worked best.
 
Bart Winters and Virco colleagues compared the vircoTYPE phenotypic score with Tibotec's darunavir mutation score, the Stanford genotype-based algorithm, and the protease inhibitor (PI) mutation list in the US product label for darunavir. They correlated each score with 8- and 24-week outcomes in the POWER trials, accounting for background antiretroviral activity by a continuous phenotypic sensitivity score.
 
(The Tibotec score includes 11 mutations: 11I, 32I, 33F, 47V, 50V, 54L, 54M, 73S, 76V, 84V, and 89V. The US label includes any mutation at protease position 30, 32, 36, 46, 47, 48, 50, 53, 54, 73, 82, 84, 88, or 90.)
 
The vircoTYPE-predicted fold change in susceptibility correlated best with week 8 virologic response for 319 POWER participants taking darunavir/ritonavir at an accuracy score of 87.4% versus 83.7% for Tibotec's darunavir mutation score, 80% for the Stanford score, and 80% for the US label mutation list. Spearman correlation (R value) for each score versus 8-week response was highest for vircoTYPE (0.46), followed by the Tibotec score (0.36), the Stanford system (0.21), and the US label mutation list (0.15).
 
Looking at 24-week responses, Winters again found the best predictive accuracy with vircoTYPE (81.8%), followed by the Tibotec mutation score (78.4%), the Stanford algorithm (76.5%), and the US label mutation list (74.6%). Spearman R values ranged from 0.45 with vircoTYPE, to 0.33 with the Tibotec score, 0.24 with the Stanford algorithm, and 0.17 with the US mutation list.
 
The number of correct response calls at week 8 ranged from 255 out of 319 (80%) for the US mutation list of 278 (87%) for vircoTYPE. Correct week 24 calls ranged from 237 (74%) with the US mutation list to 260 (81.5%) for vircoTYPE.
 
The Virco researchers proposed that vircoTYPE may perform better than genotypic scores in predicting virologic response in POWER because of "its ability to factor complex mutational profiles in its resistance interpretation and more accurately predict expected activity of darunavir against viruses with varying degrees of resistance." They cautioned that all these response predictions relate only to people enrolled in the POWER trials. The predictive power of any system could change in different populations, especially in patients who differ substantially from those in the POWER studies.
 
Speaking at the Workshop, Stanford University's Robert Shafer (the brain and brawn behind the Stanford algorithm) noted that US guidelines do not specify genotyping or phenotyping when they recommend resistance testing. European guidelines do, generally favoring genotyping for most patients but suggesting phenotyping may have an advantage in people with complex mutation patterns or in those starting a newly introduced drug. Both conditions apply to darunavir as used in the POWER trials and will for many people now starting this PI.
 
Reference
1. Winters B, Van Der Borght K, Van Craenenbroeck E, et al. Accuracy and correlation of response predictions based on 4 methodologies to interpret darunavir resistance in patients receiving darunavir containing regimens in the POWER studies. 5th European HIV Drug Resistance Workshop. March 28-30, 2007. Cascais, Portugal. Abstract 72.