Jules Levin, NATAP

Mike Saag reported the data from this study of 71 individuals. The cohort or group of patients in this study was treatment-experienced and their treatment history was verified going back to their initial regimen. There viral load was >5000 copies/ml at entry into this study, and they had to be changing therapy at time of enrollment. The baseline regimen referred to in this report is the new regimen they changed to. Plasma collected at baseline and every 3 months, and clinical and lab evaluations were taken every 3 month. Data for this study was collected prospectively.

The ViroLogic phenotypic test was used in this study and was done retrospectively on stored samples for the baseline. Because there is debate about what is the significant cut-off for resistance they analyzed it both ways: < 2.5 fold and <4 fold and found no differences whether using one or the other. Successful therapy is defined in this study as >0.5 log viral load reduction from baseline that was sustained at all follow-up visits. They also analyzed the data using a 1-log drop and didn’t see any difference in the results.

Baseline median viral load and CD4 were 70,664 copies/ml and 142. 35% were non- caucasion and 15% were female. Participants had experienced a median of 5 prior treatment regimens. 72% had prior PI use and 90% were NNRTI-naïve. In their new regimen 96% used a PI. Median duration of study follow-up was 15 months and median time to failure was 171 days.

Saag reported that using a univariate analysis looking for time to failure only 3 factors had independent prognostic value. Analysis using both cut-offs of 2.5-fold or 4-fold gave the same results:

1. Prior PI-experience
2. The number of medications in the regimen that the person was susceptible to using the phenotypic test
3. The number of medications in the regimen not used in person’s prior therapies. That is predicted by their history not to be sensitive.

The factors looked at that were not predictive in this analysis were:

Baseline viral load

Baseline CD4

Number of prior regimens

NNRTI experience

Number of medications in the regimen

PI included in regimen

When putting all these factors into a multi-variate analysis the only one that became independently predictive was the number of drugs the person was susceptible to using the phenotypic test. If susceptibility was dropped from the model then the number of drugs in the regimen (treatment history) not used previously became an important predictor, and prior PI use became a significant predictor of failure.

Using a Kaplan-Meier analysis, if the person had 2 or less drugs that they were susceptible to by phenotypic testing in their new regimen their time to failure was much more rapid than if they were sensitive to 3 or more drugs in the new regimen. This was highly statistically significant. If you look at the 1-year point, half the patients were successful with treatment (0.5 log reduction or more) when they were sensitive to 3 or more drugs at baseline. While only 18% were still successful at 1 year if they were sensitive to 2 or less drugs in the new regimen.

Saag summarized by saying that testing may be helpful in determining the next best regimen in this population. He said we’re going to have to fine-tune the phenotypic testing by identifying specific cut-offs for resistance for particular drugs. Currently, resistance testing uses generic cut-offs of resistance such as 4-fold for or 10-fold. But what if resistance to a particular drug requires 15-fold decreased susceptibility?

Saag said as studies are carried out to confirm the clinical outcome of using resistance testing, these tests can be helpful in a highly experienced group so that we don’t eliminate drugs that are still active simply because they were in a previous regimen. And similarly, so that we don’t use drugs in a regimen just because they were not used previously but to which a person may be resistant.