icon-folder.gif   Conference Reports for NATAP  
  49th ICAAC
San Francisco, CA
September 12-15, 2009
Back grey_arrow_rt.gif
Weighted Antiretroviral Sensitivity Score May Be
Better Predictor of Short-Term Response

  49th ICAAC (Interscience Conference on Antimicrobial Agents and Chemotherapy), September 12-15, 2009, San Francisco
Mark Mascolini
A weighted phenotypic sensitivity score (wPSS)--which accounts for an antiretroviral's innate activity against susceptible and resistant HIV strains--predicted short-term virologic response better than an unweighted phenotypic sensitivity score (PSS) or any of three genotypic sensitivity scores (GSS) in a study by Virco scientists [1]. All sensitivity scores sum the activity of drugs in a regimen. Weighting has emerged as a potentially more precise way to estimate a regimen's activity because individual antiretrovirals differ in how strongly they suppress resistant and nonresistant virus.
Virco investigators considered 5426 treatment change episodes from clinical trials and cohort studies to develop weights for current antiretrovirals. They used vircoTYPE clinical cutoffs for the phenotypic sensitivity (fold change in susceptibility) part of the wPSS equation [2]. Then they compared the accuracy of wPSS with the accuracy of standard PSS and three GSS predictions of short-term response after a treatment change: at least a 1-log (10-fold) drop in viral load 8 weeks after the new regimen began or an undetectable viral load after 24 weeks. The study populations included patients with no mutations and patients with mutations classified as "experienced" or "salvage" by a discriminant analysis trained on the genotypes of patients in clinical trials.
The researchers determined weights by estimation modeling of week-8 change in viral load. The Virco team believes their approach can "cope with imbalances in the dataset with respect to other patients (baseline viral load), virus (resistance) or treatment characteristics (other drugs classes in the regimen) that may affect virologic response." They figured individual PI weights by comparing virologic response at week 8 for particular PIs compared with lopinavir, which they assigned a weight of 1.0.
Weights for nucleosides were 0.3 for 1 active nucleoside in a regimen and 1.1 for 2 or more active nucleosides. All nonnucleosides carried a weight of 1.0. Weights for protease inhibitors were 0.54 for nelfinavir, 0.64 for tipranavir/ritonavir, 0.69 for fosamprenavir/ritonavir, 0.74 for indinavir/ritonavir, 1.00 for atazanavir/ritonavir, 1.00 for lopinavir/ritonavir, to 1.25 for darunavir/ritonavir 800 mg once daily, and 1.57 for darunavir/ritonavir 600 mg twice daily.
In a validation set of 1923 treatment change episodes, receiver operating characteristic (ROC) curve analysis determined the following rates for accurate prediction of 8-week virologic response for all regimens considered:
· wPSS: 82%
· Standard PSS: 78%
· ANRS GSS: 76%
· HIVdb (Stanford) GSS: 76%
· Rega GSS: 78%
ROC curve analysis showed that weighted PSS also outperformed the other methods in predicting 8-week 1-log drop in viral load and 24-week undetectable response rates to salvage regimens (with dropouts considered failures in the 24-week analysis):
· wPSS: 79% at week 8, 79% at week 24
· Standard PSS: 71% at week 8, 67% at week 24
· ANRS GSS: 69% week 8, 64% week 24
· HIVdb GSS: 66% week 8, 62% week 24
· Rega GSS: 72% week 8, 70% week 24
At this point the weighted scoring system does not include antiretrovirals in new classes-entry inhibitors and integrase inhibitors.
The Virco investigators conclude that wPSS is "a more robust predictor of response at week 24 than an unweighted PSS and several GSS." They note that wPSS remained highly accurate in salvage patients, when the other scoring systems lost accuracy.
1. Winters B, Lantican L, Van Walle I, et al. Development of a potency weighted sensitivity score to assess the activity of antiretroviral drug regimens. 49th ICAAC (Interscience Conference on Antimicrobial Agents and Chemotherapy). September 12-15, 2009. San Francisco. Abstract H-893.
2. Winters B, Montaner J, Harrington R, et al. Determination of clinically relevant cutoffs for HIV-1 phenotypic resistance estimates through a combined analysis of clinical trials and cohort data. JAIDS. 2008;48:26-34.