Report From First of Two Day FDA Hearing on HIV Drug Resistance 

Nov 2
Gaithersburg, MD at Holiday Inn 

This report consists of selected key points that appear to be potentially important regarding the use of resistance testing. The points are reported in a free flow manner. 

In GART Study which used genotypic resistance testing--

Those with 90 mutation had .30 log increase and those with 30 at baseline had .41 log reduction.

GART patients (individuals who used genotypic TruGene Visible Genetics GenotypicTest) were more likely to receive more active number of drugs. Those with greater number of active drugs had better VL reduction

Those with 3 or more sensitive drugs did better than those sensitive to less than 3 drugs. For each additional drug a person was sensitive to the response improved incrementally. Those sensitive to 3 drugs did best. Those failing only 1 PI from previous therapy may be helped more with resistance testing than if person had previously failed multiple PIs.

In ACTG 333 it was found that-

lower risk of progression per log reduction in RNA

baseline RNA was highly predictive of treatment failure

failure risk increases with number of baseline PI mutations. For each additional PI mutation there was increased 2-fold risk of failure. The more PI mutations the less likely to achieve viral suppression.

There was a trend associating the number of PI mutations and baseline HIV-RNA 
and CD4.

In another presentation of an analysis researcher found-

baseline HIV-RNA correlates with virologic failure
the number of NRTI & NNRTI mutations is associated with virologic failure

Although a few said 4-fold phenotypic cut-off is better predictor of virologic response than 10-fold cut-off 

Doug Richman said he would prefer to select drugs for individuals with 4-fold as cutoff.

Mellors said cut-off for future PIs may be 30 fold. I assume he's referring to tipranavir and ABT-378. Scott Hammer said we're unsure of the clinical effect of higher in vitro break points or cut-offs of resistance for individual drugs. The committee discussed who should be responsible for establishing these cut-offs.

Veronica Miller said phenotypic resistance was significant predictor in her study and either 4 or 10 fold was good predictor.

Doug Mayers said key is developing cut-offs for different drugs. Another consideration is failure can be related to tolerability and thus adherence but this can get confused with resistance developing. Only 3TC resistance may be detectable because the patient stopped taking indinavir consistently (non-adherence).

Mellors listed a number of prospective studies ongoing or planned looking at resistance testing and its utility

Mellors reported 12 week preliminary data from study VIRA 3001--Experienced pts- >2 NRTIs and 1 PI; N=127

Using a "modified ITT analysis" (observed data) he found a statistically significant difference in viral load response of .5 log or greater at each time point between those using resistance test and those not using it. The percent undetectable at 16 wks was 52% (or 62%; I'm not sure) vs 33% in control arm.

Mellors said he feels studies have consistently established an association between genotypic test & phenotypic test & virological response consistent and the data supports the utility of resistance testing.

FDA Statistician's data suggest virologic failure associated with geno and pheno measures at baseline but additional studies neded to confirm this.

80% or more mutations and their association with resistance for specific drugs was agreed to by general consensus of resistance researchers. Rich D'Aquila said we need data bases and these databases are being establiahed. These databases will accumulate data on patient samples and their correlation of viral load responses, outcome and baseline mutations. Virco has started such a database which they call "Virtual Phenotyping" and Visible Genetics is planning such a database. Other groups are considering such databases.

Committee discussed limitations of resistance testing-

Utility of resistance testing is associated with capacity to interpret testing test results, patient's treatment experiences, and other consideration

Carla Pertinelli with the NIH said we need more longer term data from prospective studies correlating virological outcome with resistance testing.

Hanry Mazur, Roger Pomerantz and Mark Harrington (community representative) raised issue of clinical benefit and its association with virological response. Virological response may not always predict longer term clinical benefit when considering such things as quality of life and other long term benefits. Hammer concluded you can't manage a person's HIV situation from a person's resistance profile. Another committee member said we need to look at long term larger studies to better characterize the utility of viral load reductions and resistance testing.

Scott Hammer mistakenly said two times "clinical income" instead of clinical outcome. Joke.

New resistance data such as the emergence of new mutations will emerge in future. For example the 69 mutation was recently discovered. 

Adherence, blood levels, absorption, and other considerations should be monitored in addition to correlating outcome with resistance testing. Drug companies and others conducting clinical studies using resistance testing should also evaluate and report these measures.

Hammer said tests are here and are being used but prospective studies are needed to help better characterize resistance testing utility.

Heidi Jolson and Scott Hammer agreed additional meetings are necessary to improve understanding utility and application of resistance testing